# Data MCP Directory: every page as markdown The Data MCP Directory lists MCP servers for data tools (warehouses, dbt, ingestion connectors, BI tools, catalogs and observability tools), each read from its own repository or documentation with the version and the read date. It lists and does not rank. --- # MCP servers for data tools, checked one by one DIRECTORY, checked 10 Nov 2026 **The short answer.** The Data MCP Directory lists 30 MCP servers for 26 data tools, from Airbyte to ThoughtSpot, each read from its own repository or documentation on 10 Nov 2026. 26 are published by the tool's vendor, 3 by a vendor's Labs organization and 1 by the community. 15 run hosted, 3 run locally and 12 offer both. Filed by data tool, then by server name. Kind and transport are labels read from each source; nothing is ordered by merit. | Data tool | Server | Kind | Transport | Access | Tools | Version | Last release | | --- | --- | --- | --- | --- | --- | --- | --- | | Airbyte | [Airbyte MCP (hosted)](/servers/airbyte-mcp/) | Official | Remote (hosted) | Configurable | 50 | Hosted service (no versioned release); PyAirbyte package 0.71.3 ships an experimental local airbyte.mcp server | none found | | Alation | [Alation AI Agent SDK MCP server](/servers/alation-mcp/) | Official | Remote and local | Not stated in the source read | 12 | 1.0.0 | 26 Oct 2026 | | Amazon Redshift | [Amazon Redshift MCP Server (AWS Labs)](/servers/awslabs-redshift-mcp-server/) | Official, Labs org | Local (stdio) | Configurable | 8 | 0.1.3 | 9 Nov 2026 | | Atlan | [Atlan MCP Server](/servers/atlan-mcp/) | Official | Remote (hosted) | Configurable | 41 | Hosted at mcp.atlan.com/mcp (unversioned); local atlan-mcp-server 0.3.3 (deprecated) | 17 Feb 2026 | | BigQuery | [BigQuery remote MCP server](/servers/bigquery-remote-mcp/) | Official | Remote (hosted) | Configurable | 9 | Docs reference last updated 3 Nov 2026 | none found | | BigQuery | [MCP Toolbox for Databases (BigQuery tools)](/servers/mcp-toolbox-bigquery/) | Official | Remote and local | Configurable | 9 | V1.16.0 | 5 Nov 2026 | | Collibra | [Collibra MCP Server (chip)](/servers/collibra-mcp/) | Official | Remote and local | Configurable | 49 | V0.0.67 | 27 Oct 2026 | | Cube | [Cube MCP server](/servers/cube-mcp/) | Official | Remote (hosted) | Read and write | 30 | Hosted service (no versioned release) | none found | | Databricks | [Databricks Labs MCP servers](/servers/databrickslabs-mcp/) | Official, Labs org | Remote and local | Not stated in the source read | 3 | none found on 10 Nov 2026 | none found | | Databricks | [Databricks managed MCP servers](/servers/databricks-managed-mcp/) | Official | Remote (hosted) | Configurable | 4 | Docs page last updated Nov 9, 2026 | none found | | DataHub | [DataHub MCP Server](/servers/datahub-mcp/) | Official | Remote and local | Configurable | 22 | 0.8.0 | 21 Oct 2026 | | dbt | [dbt MCP Server](/servers/dbt-mcp/) | Official | Remote and local | Configurable | 52 | 2.10.1 | 3 Nov 2026 | | Euno | [Euno MCP](/servers/euno-mcp/) | Official | Remote (hosted) | Configurable | 56 | Hosted service at api.app.euno.ai/mcp, no version published (docs read 10 Nov 2026) | none found | | Fivetran | [Fivetran MCP Server](/servers/fivetran-mcp/) | Official | Remote and local | Configurable | 53 | V0.5.1 | 6 Oct 2026 | | Hex | [Hex MCP server](/servers/hex-mcp/) | Official | Remote (hosted) | Read and write | 18 | Hosted service (no versioned release) | none found | | Looker | [Looker tools in MCP Toolbox for Databases](/servers/looker-mcp-toolbox/) | Official | Remote and local | Configurable | 55 | V1.16.0 | 5 Nov 2026 | | Monte Carlo | [Monte Carlo MCP Server](/servers/monte-carlo-mcp/) | Official | Remote (hosted) | Configurable | 15 | Hosted at mcp.getmontecarlo.com/mcp, no version published | none found | | Omni | [Omni MCP Server](/servers/omni-mcp/) | Official | Remote (hosted) | Configurable | 26 | Hosted service (no versioned release) | none found | | OpenMetadata | [OpenMetadata MCP Server](/servers/openmetadata-mcp/) | Official | Remote (hosted) | Read and write | 15 | 2.0.7 (bundled with platform release 2.0.7-release; 2.1.0-rc1 also listed) | 27 Oct 2026 | | Power BI | [Power BI Modeling MCP Server](/servers/powerbi-modeling-mcp/) | Official | Remote and local | Configurable | 21 | 1.0.1 | 6 Nov 2026 | | Secoda | [Secoda MCP](/servers/secoda-mcp/) | Official | Remote and local | Not stated in the source read | 8 | No releases; main branch last commit 26 Jan 2026 | none found | | Select Star | [Select Star MCP Server](/servers/select-star-mcp/) | Official | Remote (hosted) | Read only | 7 | Hosted, no version published | none found | | Sifflet | [Sifflet MCP Server](/servers/sifflet-mcp/) | Official | Local (stdio) | Read and write | 10 | 0.1.7 (PyPI); main at commit 1f390ca | 3 Aug 2026 | | Sigma | [Sigma MCP server](/servers/sigma-mcp/) | Official | Remote (hosted) | Read and write | 0 | Hosted service (no versioned release) | none found | | Snowflake | [mcp-snowflake-server (community)](/servers/isaacwasserman-mcp-snowflake-server/) | Community | Local (stdio) | Configurable | 8 | 0.4.0 | 10 Apr 2025 | | Snowflake | [Snowflake Labs MCP Server](/servers/snowflake-labs-mcp/) | Official, Labs org | Remote and local | Configurable | 6 | 1.4.2 | 15 May 2026 | | Snowflake | [Snowflake-managed MCP server](/servers/snowflake-managed-mcp/) | Official | Remote (hosted) | Configurable | 6 | Docs page, MCP revision 25 Nov 2025 | none found | | SYNQ | [SYNQ MCP (Coalesce Quality)](/servers/synq-mcp/) | Official | Remote (hosted) | Configurable | 134 | Hosted, no version published | none found | | Tableau | [Tableau MCP](/servers/tableau-mcp/) | Official | Remote and local | Configurable | 40 | 6.0.4 | 9 Nov 2026 | | ThoughtSpot | [ThoughtSpot MCP Server](/servers/thoughtspot-mcp-server/) | Official | Remote (hosted) | Read and write | 20 | 0.5.0 (npm and package.json); main at commit 820bf16f6aa3 | 6 Jun 2025 | 30 MCP servers for 26 data tools, each read from its own repository or documentation on 10 Nov 2026. ## Six kinds of data tool ### 4 data tools: Warehouses (8) 6 remote, 5 local 8 servers for Amazon Redshift, BigQuery, Databricks, Snowflake. [Open warehouses](/categories/warehouses/) ### 1 data tool: Transformation (1) 1 remote, 1 local 1 server for dbt. [Open transformation](/categories/transformation/) ### 2 data tools: Ingestion connectors (2) 2 remote, 1 local 2 servers for Airbyte, Fivetran. [Open ingestion connectors](/categories/ingestion/) ### 8 data tools: BI and analytics (8) 8 remote, 3 local 8 servers for Cube, Hex, Looker, Omni, Power BI, Sigma, Tableau, ThoughtSpot. [Open bi and analytics](/categories/bi/) ### 8 data tools: Catalogs and metadata (8) 8 remote, 4 local 8 servers for Alation, Atlan, Collibra, DataHub, Euno, OpenMetadata, Secoda, Select Star. [Open catalogs and metadata](/categories/catalogs/) ### 3 data tools: Data observability (3) 2 remote, 1 local 3 servers for Monte Carlo, Sifflet, SYNQ. [Open data observability](/categories/observability/) ## What each server is for, in one sentence - [Airbyte MCP (hosted)](/servers/airbyte-mcp/) (Airbyte): Airbyte's hosted server lets an agent build, run and troubleshoot pipelines as the signed-in user, with a documented read-only mode; it serves pipeline and catalog state. - [Alation AI Agent SDK MCP server](/servers/alation-mcp/) (Alation): Alation's server offers a lineage tool and a catalog search agent, and its remote form hides tools the current user cannot use. - [Amazon Redshift MCP Server (AWS Labs)](/servers/awslabs-redshift-mcp-server/) (Amazon Redshift): The AWS Labs server is read-only by default with an explicit read-write opt-in, confirmation prompts and metadata discovery over the Redshift Data API. - [Atlan MCP Server](/servers/atlan-mcp/) (Atlan): Atlan's hosted per-tenant server runs as the signed-in user, scopes context by persona, offers an optional read-only mode and proposes writes for approval. - [BigQuery remote MCP server](/servers/bigquery-remote-mcp/) (BigQuery): Google Cloud's managed endpoint gives IAM-scoped access to a SELECT-only query tool and dataset and table metadata tools. - [MCP Toolbox for Databases (BigQuery tools)](/servers/mcp-toolbox-bigquery/) (BigQuery): Google's open-source Toolbox offers BigQuery SQL and metadata tools with a read-only mode, dataset allowlists and Knowledge Catalog search. - [Collibra MCP Server (chip)](/servers/collibra-mcp/) (Collibra): Collibra's server covers lineage with transformation logic, business terms and column semantics, classification, stewardship edits and data quality, under the connecting user's permissions. - [Cube MCP server](/servers/cube-mcp/) (Cube): Cube's hosted server searches the semantic layer, runs SQL with row-level security as the signed-in user and edits the semantic model on a branch. - [Databricks Labs MCP servers](/servers/databrickslabs-mcp/) (Databricks): The Labs server exposes Unity Catalog functions, vector search indexes and Genie spaces as tools; its README marks it deprecated in favor of the managed servers. - [Databricks managed MCP servers](/servers/databricks-managed-mcp/) (Databricks): Databricks-provided MCP endpoints run under the caller's Unity Catalog grants and include Genie spaces for questions over a governed semantic layer. - [DataHub MCP Server](/servers/datahub-mcp/) (DataHub): DataHub's server returns column-level lineage with paths and SQL transformations, plus glossary, domain and owner discovery, with per-user tokens in HTTP mode. - [dbt MCP Server](/servers/dbt-mcp/) (dbt): dbt Labs' server combines the dbt CLI, the Discovery API with model and column lineage, the Semantic Layer and SQL tools on dbt Platform. - [Euno MCP](/servers/euno-mcp/) (Euno): Euno's hosted OAuth endpoint serves persona-scoped lineage, impact analysis, EQL search and catalog tools over its metadata graph, with no warehouse query tool. - [Fivetran MCP Server](/servers/fivetran-mcp/) (Fivetran): Fivetran's server mirrors its REST API behind a read-by-default scope with a deny-list, so an agent can inspect connections, schemas and sync status. - [Hex MCP server](/servers/hex-mcp/) (Hex): Hex's hosted server runs Threads (agent analyses over the user's data connections) and notebook editing under workspace roles and project permissions. - [Looker tools in MCP Toolbox for Databases](/servers/looker-mcp-toolbox/) (Looker): Google's Toolbox exposes Looker models, explores, dimensions, measures and LookML project files, plus instance health and usage tools. - [Monte Carlo MCP Server](/servers/monte-carlo-mcp/) (Monte Carlo): Monte Carlo's hosted observability server uses per-user OAuth or scoped keys, a per-request read-only header and tool allow-lists. - [Omni MCP Server](/servers/omni-mcp/) (Omni): Omni's hosted server runs as a specific Omni user and exposes models, topics, saved queries and apps, so answers follow the semantic model and the user's permissions. - [OpenMetadata MCP Server](/servers/openmetadata-mcp/) (OpenMetadata): OpenMetadata's built-in server offers per-user OAuth, persona-scoped context, table and optional column lineage, and write tools for lineage, entities and tests. - [Power BI Modeling MCP Server](/servers/powerbi-modeling-mcp/) (Power BI): Microsoft's authoring server reads and writes semantic-model objects (tables, columns, DAX measures, relationships, RLS roles, TMDL) under the user's Fabric permissions, with a read-only switch. - [Secoda MCP](/servers/secoda-mcp/) (Secoda): Secoda's server combines catalog search, glossary, lineage and SQL over a workspace, as a hosted endpoint and an open-source wrapper. - [Select Star MCP Server](/servers/select-star-mcp/) (Select Star): Select Star's hosted, read-only server exposes catalog search, lineage, query history and semantic models with token setup. - [Sifflet MCP Server](/servers/sifflet-mcp/) (Sifflet): Sifflet's local server covers asset search, incidents, monitors as code and downstream impact from one Sifflet API token. - [Sigma MCP server](/servers/sigma-mcp/) (Sigma): Sigma's hosted server searches data models and workbooks, reads column definitions, and queries and builds workbooks under the user's own Sigma permissions. - [mcp-snowflake-server (community)](/servers/isaacwasserman-mcp-snowflake-server/) (Snowflake): A community server for schema browsing and read-only SQL on Snowflake, with write tools off by default. - [Snowflake Labs MCP Server](/servers/snowflake-labs-mcp/) (Snowflake): The Labs server gives config-driven access to Cortex services, object management and semantic views under the connecting role; its README marks it deprecated. - [Snowflake-managed MCP server](/servers/snowflake-managed-mcp/) (Snowflake): Snowflake's hosted endpoint applies RBAC per server and per tool with OAuth, and serves Cortex Analyst over semantic views, Cortex Search and SQL. - [SYNQ MCP (Coalesce Quality)](/servers/synq-mcp/) (SYNQ): SYNQ's hosted observability and lineage server uses per-user OAuth with tiered consent scopes, and tools hide when a scope is not granted. - [Tableau MCP](/servers/tableau-mcp/) (Tableau): Tableau's server exposes data source metadata with calculation strings and Pulse metric definitions per signed-in user, plus, on Tableau+, a knowledge tool for lineage and impact. - [ThoughtSpot MCP Server](/servers/thoughtspot-mcp-server/) (ThoughtSpot): ThoughtSpot's server returns Spotter answers, searches objects with owner, tag and verified filters, and creates dashboards and models within the user's privileges. ## By situation - [A data platform lead whose analysts no longer trust the dashboards](/situations/untrusted-dashboards/): Check lineage, impact and incident tools across catalogs, observability and BI servers. - [An analytics engineering team on dbt and Snowflake or Databricks](/situations/dbt-warehouse/): Check the dbt, Snowflake and Databricks servers side by side: what runs SQL, what reads models. - [A company putting AI agents on its data this year](/situations/agents-this-year/): Check transport, auth as the user, write tools and tool-definition size across every server. - [A governance lead who must show where a regulated field flows](/situations/regulated-field/): Check lineage depth, classification tools, audit and scoping in catalog and warehouse servers. - [A BI team with thousands of reports and no map of what feeds them](/situations/bi-sprawl/): Check what each BI server exposes about data sources, semantic models and lineage. ## What an MCP server for a data tool is An MCP server is a program that speaks the Model Context Protocol, so an agent client such as Claude, Cursor or ChatGPT can call the tools it lists. For a data tool, those tools run SQL, read metadata, follow lineage, query a semantic layer, or change something: a pipeline, a dashboard, a catalog entry. The tools differ more than the protocol does. Some servers are hosted by the vendor and sign the user in with OAuth; others run on the engineer's laptop over stdio with a token in an environment variable. Some only read; others write and ask for confirmation first. This directory records those facts for each server, from its own source, so an engineer can write down what a server really does before wiring an agent to it. Data catalogs, lineage tools and context platforms as categories are covered elsewhere; this directory covers the MCP servers themselves. ## Guides and glossary - [What an MCP server for data should expose](/guides/what-an-mcp-server-for-data-should-expose/) - [Read versus write on data MCP servers](/guides/read-versus-write/) - [Scoping context to the person asking](/guides/scoping-context-to-the-person-asking/) - [Why tool definition size matters](/guides/why-tool-definition-size-matters/) - [Auth and audit on data MCP servers](/guides/auth-and-audit/) - [Remote versus local MCP servers for data](/guides/remote-versus-local-servers/) - [The glossary, 36 terms](/glossary/) - [MCP server news, 12 dated items](/news/) ## Sources - Data MCP Directory datasets (CSV and JSON): https://datamcpservers.com/datasets/ (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/ --- # The directory lists and does not rank: each entry is read from its own source and dated METHOD, read 10 Nov 2026 **The short answer.** Each of the 30 entries is read from the server's own repository or documentation, with the version or commit and the read date, 10 Nov 2026. Every server gets the same fourteen fields in the same order. The directory lists and does not rank: entries are filed alphabetically by data tool, then by server name. The Data MCP Directory lists and does not rank. It records what each MCP server for a data tool publishes about itself and what can be observed in its source, field by field, and stops there. No page carries a total, an order of merit or a rating, and no page names a choice. Every entry is read from the server's own repository, package registry page or vendor documentation. The version or commit read and the date of reading go with every entry. A server whose own source could not be read is not listed: on 10 Nov 2026 that rule admitted 30 servers. ## The fourteen fields, in the order every entry shows them | Field | What it records | | --- | --- | | Data tool | The product the server connects an agent to, such as Snowflake or Tableau. | | Publisher | Who publishes the server, as its repository or documentation names them. | | Kind | Official (the data tool's vendor), Official, Labs org (a vendor's Labs organization) or Community. | | Repository or docs | The page the entry was read from. | | Version or commit | The version, tag or commit read; for hosted services without versions, what the docs give. | | Last release | The date of the latest release found, or none found on the read date. | | Transport | Remote (a hosted endpoint over HTTP), Local (stdio on the user's machine), or both. | | Auth | How the server authenticates, as documented: OAuth as the user, tokens, keys or service accounts. | | Access | Read only, read and write, configurable, or not stated in the source read. | | Tools listed | How many tool names the evidence records. | | Tool definitions read at | Where the tool names and descriptions were read. | | Context weight | Tokens in the published tool names and descriptions, from the Context Weight Index, or why it could not be measured. | | Status | Whether it exists under that name, deprecations, renames and moves, as read. | | Read | The date the entry was read. | ## How kind, transport and access are labelled - **Official:** The data tool's own vendor publishes the server, in its own repository or documentation. - **Official, Labs org:** Published under a vendor's Labs organization on GitHub (AWS Labs, Snowflake Labs, Databricks Labs); the label says so because Labs projects can carry different support terms. - **Community:** Published by an individual or a group other than the vendor. - **Remote, Local, Both:** Remote means a hosted endpoint the client reaches over HTTP; local means the client starts the server as a process over stdio; both means the source documents both. - **Access:** Taken from the source: read only where the server documents only read tools or a read-only default; configurable where it documents switches, scopes or modes; read and write where write tools ship without a documented switch. ## What counts as evidence Every fact on the site carries a source URL and the date it was read. A publisher's statement about its own server is written as that publisher's claim. Where something was looked for and not found, the entry says none found on 10 Nov 2026, which records what the sources read showed on that date and is never written as a missing feature. Facts are grouped on each entry by topic: publisher and releases; clients, auth and scoping; access and safety; what it reaches; lineage and impact; definitions, ownership and rules; freshness and upkeep. The topics are a filing system, and every entry uses the same one. | Kind of fact | Facts | What it means | | --- | --- | --- | | Vendor claim | 413 | The publisher's own documentation or README says it; written as the publisher's claim. | | Observed | 56 | Read directly in the source code, the tool list, a package registry or the release page. | | Third party | 1 | A source other than the publisher, such as a registry or another vendor. | | None found | 52 | Looked for in the sources read and not found on 10 Nov 2026; never written as a missing feature. | 522 facts across 30 servers, read 10 Nov 2026. ## Why nothing is ordered by merit Lists are filed alphabetically by data tool, then by server name, on every page and in every download. Kind and transport are labels a reader can filter on, not an order. Side-by-side pages put two servers' facts in the same rows and end without a conclusion. The one sortable table on the site is the Context Weight Index, which is a measurement: it shows how many tokens each server's published tool definitions take, measured one way for all, and a reader can sort it by that number. ## What the directory does not cover The directory does not test servers against live accounts. It reads what each publisher documents and what the source shows, and says which. MCP gateways, agent identity and agent governance platforms are outside its subject. So are data catalogs, lineage tools and context platforms as product categories: their servers are listed here, and the products themselves are not reviewed. ## Sources - The directory, CSV: https://datamcpservers.com/datasets/data-mcp-directory.csv (read 10 Nov 2026) - The evidence, CSV: https://datamcpservers.com/datasets/data-mcp-evidence.csv (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/method/ --- # Airbyte MCP (hosted): 50 tools, remote (hosted), from Airbyte SERVER, read 10 Nov 2026 **The short answer.** Airbyte MCP (hosted) is an official MCP server for Airbyte, published by Airbyte and read from its own source on 10 Nov 2026. It runs remote (hosted), and 50 tools are named. Version read: Hosted service (no versioned release); PyAirbyte package 0.71.3 ships an experimental local airbyte.mcp server. - **Data tool:** [Airbyte](/mcp/airbyte/) - **Publisher:** Airbyte - **Kind:** Official - **Repository or docs:** [docs.airbyte.com/platform/airbyte-mcp](https://docs.airbyte.com/platform/airbyte-mcp) - **Version or commit:** Hosted service (no versioned release); PyAirbyte package 0.71.3 ships an experimental local airbyte.mcp server - **Last release:** none found on 10 Nov 2026 - **Transport:** Remote (hosted) - **Auth:** OAuth (Streamable HTTP) as the signed-in Airbyte user, or an Airbyte application for non-browser clients; tools run with that user's Airbyte role. - **Access:** Configurable - **Tools listed:** 50 named in the evidence - **Tool definitions read at:** [docs.airbyte.com/platform/airbyte-mcp/tools.md](https://docs.airbyte.com/platform/airbyte-mcp/tools.md) - **Context weight:** 870 tokens for 50 tool definitions (50 with a description), measured from the docs page read - **Status:** Exists on 10 Nov 2026 at mcp.airbyte.com/mcp for Airbyte Cloud and Flex; self-managed Core can run it locally with known issues. A separate PyAirbyte (airbyte.mcp) server and an airbyte-agent-connectors SDK also exist; this file covers the documented Airbyte MCP. - **Read:** 10 Nov 2026 ## In one sentence Airbyte's hosted server lets an agent build, run and troubleshoot pipelines as the signed-in user, with a documented read-only mode; it serves pipeline and catalog state. ## What Airbyte MCP (hosted) exposes 50 tools are named in the evidence read on 10 Nov 2026. `get_default_workspace_context`, `list_organizations`, `describe_organization`, `get_organization_billing_status`, `list_workspaces`, `describe_workspace`, `set_default_workspace`, `list_connectors`, `describe_connector`, `deploy_connector`, `deploy_noop_destination`, `check_connector`, `rename_connector`, `update_connector_config`, `permanently_delete_connector`, `list_connections`, `describe_connection`, `create_connection`, `update_connection`, `rename_connection`, `set_connection_selected_streams`, `set_connection_table_prefix`, `discover_source_catalog`, `get_connection_artifact`, `permanently_delete_connection`, `run_connection_sync`, `cancel_connection_sync`, `get_connection_sync_status`, `list_connection_sync_jobs`, `get_connection_sync_logs`, `show_workspace_sync_status`, `show_connection_sync_history`, `show_connector_definitions`, `list_custom_source_definitions`, `get_custom_source_definition`, `get_connector_builder_draft_manifest`, `publish_custom_source_definition`, `update_custom_source_definition`, `permanently_delete_custom_source_definition`, `get_external_access_skill_docs`, `execute_external_api_query`, `execute_external_sql_query`, `execute_external_search_query`, `get_search_index_status`, `request_search_index_backfill`, `list_connector_definitions`, `get_connector_definition`, `get_connector_definition_version_history`, `get_api_docs_urls`, `search_airbyte_knowledge_sources` ## Not found in the sources read - Any lineage tool in the Airbyte MCP tools page (column, transformation or BI lineage). [Source](https://docs.airbyte.com/platform/airbyte-mcp/tools.md). - Glossary, ownership or classification tools in the Airbyte MCP tools page. [Source](https://docs.airbyte.com/platform/airbyte-mcp/tools.md). - Scheduled or event-driven metadata collection described in the Airbyte MCP docs; tools query on demand. [Source](https://docs.airbyte.com/platform/airbyte-mcp.md). ## The evidence on Airbyte MCP (hosted), by topic ### Publisher and releases - Airbyte documents the Airbyte MCP as its own Model Context Protocol server connecting agents to an Airbyte organization with the user's own permissions. (Airbyte's own claim; [source](https://docs.airbyte.com/platform/airbyte-mcp.md), read 10 Nov 2026.) - The hosted server has no versioned release; its URL is mcp.airbyte.com/mcp. (Airbyte's own claim; [source](https://docs.airbyte.com/platform/airbyte-mcp/install.md), read 10 Nov 2026.) - The airbyte package on PyPI (PyAirbyte) is at version 0.71.3, uploaded 26 Oct 2026, and includes an experimental airbyte.mcp server module. (observed in the source; [source](https://pypi.org/pypi/airbyte/json), read 10 Nov 2026.) - The PyAirbyte reference describes airbyte.mcp.server as an experimental MCP server for connector management. (Airbyte's own claim; [source](https://docs.airbyte.com/llms.txt), read 10 Nov 2026.) ### Clients, auth and scoping - The agent acts on behalf of the Airbyte user who signed in, and every tool runs with that user's Airbyte permissions. (Airbyte's own claim; [source](https://docs.airbyte.com/platform/airbyte-mcp/tools.md), read 10 Nov 2026.) - Direct reads use the credentials stored in the connector rather than the asking user, so source-side access is not per-user. (Airbyte's own claim; [source](https://docs.airbyte.com/platform/airbyte-mcp/tools.md), read 10 Nov 2026.) - With the context layer enabled, agents can read live source data, query supported destinations and search synced data by keyword, meaning or both. (Airbyte's own claim; [source](https://docs.airbyte.com/platform/airbyte-mcp.md), read 10 Nov 2026.) ### Access and safety - Read-only mode, enabled with the X-MCP-Readonly-Mode header or ?read-only=true, shows only read-only tools and hides run/cancel sync tools. (Airbyte's own claim; [source](https://docs.airbyte.com/platform/airbyte-mcp/advanced-configuration.md), read 10 Nov 2026.) - Permanent deletes only work when the resource name contains delete-me or deleteme. (Airbyte's own claim; [source](https://docs.airbyte.com/platform/airbyte-mcp/tools.md), read 10 Nov 2026.) - Headers can only hide tools and cannot grant permissions the user's Airbyte role lacks. (Airbyte's own claim; [source](https://docs.airbyte.com/platform/airbyte-mcp/advanced-configuration.md), read 10 Nov 2026.) - Direct reads and searches are read-only. (Airbyte's own claim; [source](https://docs.airbyte.com/platform/airbyte-mcp.md), read 10 Nov 2026.) ### What it reaches - Doc examples cover Postgres, BigQuery, Snowflake, HubSpot and Salesforce, i.e. ingestion sources and warehouse destinations, but no BI or transformation tools. (Airbyte's own claim; [source](https://docs.airbyte.com/platform/airbyte-mcp.md), read 10 Nov 2026.) - The hosted server works with all editions of Airbyte Cloud and Airbyte Flex; self-managed Airbyte Core can run it locally with known issues. (Airbyte's own claim; [source](https://docs.airbyte.com/platform/airbyte-mcp.md), read 10 Nov 2026.) - Clients must support remote MCP servers with OAuth and the Streamable HTTP transport. (Airbyte's own claim; [source](https://docs.airbyte.com/platform/airbyte-mcp/install.md), read 10 Nov 2026.) ### Lineage and impact - Describe_connection returns a connection's streams, schedule and status, and discover_source_catalog is a catalog tool (stream level, no column lineage documented). (Airbyte's own claim; [source](https://docs.airbyte.com/platform/airbyte-mcp/tools.md), read 10 Nov 2026.) ### Definitions, ownership and rules - Connector skills describe the entities, actions and parameters each connector supports, which is connector documentation rather than business definitions. (Airbyte's own claim; [source](https://docs.airbyte.com/platform/airbyte-mcp.md), read 10 Nov 2026.) ### Freshness and upkeep - Tools check sync status, read job logs and explain failures, so pipeline state is read live from Airbyte. (Airbyte's own claim; [source](https://docs.airbyte.com/platform/airbyte-mcp.md), read 10 Nov 2026.) ### Profile - Each organization gets 100,000 direct reads and searches per month at no cost; other MCP tools do not count toward the limit. (Airbyte's own claim; [source](https://docs.airbyte.com/platform/airbyte-mcp.md), read 10 Nov 2026.) ## Read next - [MCP servers for Airbyte](/mcp/airbyte/): 1 listed - [Airbyte MCP (hosted) and Fivetran MCP Server, side by side](/compare/airbyte-mcp-vs-fivetran-mcp/) - [The Context Weight Index](/context-weight/) ## Sources - docs.airbyte.com/platform/airbyte-mcp.md: https://docs.airbyte.com/platform/airbyte-mcp.md (read 10 Nov 2026) - docs.airbyte.com/platform/airbyte-mcp/install.md: https://docs.airbyte.com/platform/airbyte-mcp/install.md (read 10 Nov 2026) - pypi.org/pypi/airbyte/json: https://pypi.org/pypi/airbyte/json (read 10 Nov 2026) - docs.airbyte.com/llms.txt: https://docs.airbyte.com/llms.txt (read 10 Nov 2026) - docs.airbyte.com/platform/airbyte-mcp/tools.md: https://docs.airbyte.com/platform/airbyte-mcp/tools.md (read 10 Nov 2026) - docs.airbyte.com/platform/airbyte-mcp/advanced-configuration.md: https://docs.airbyte.com/platform/airbyte-mcp/advanced-configuration.md (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/servers/airbyte-mcp/ --- # Alation AI Agent SDK MCP server: 12 tools, remote and local, from Alation SERVER, read 10 Nov 2026 **The short answer.** Alation AI Agent SDK MCP server is an official MCP server for Alation, published by Alation and read from its own source on 10 Nov 2026. It runs remote and local, and 12 tools are named. Version read: 1.0.0. Last release: 26 Oct 2026. - **Data tool:** [Alation](/mcp/alation/) - **Publisher:** Alation - **Kind:** Official - **Repository or docs:** [github.com/Alation/alation-ai-agent-sdk](https://github.com/Alation/alation-ai-agent-sdk) - **Version or commit:** 1.0.0 - **Last release:** 26 Oct 2026 - **Transport:** Remote and local - **Auth:** OAuth client credentials of a service account (role-limited) or user refresh token; remote MCP at /ai/mcp needs client_id and client_secret (no dynamic client registration). - **Access:** Not stated in the source read - **Tools listed:** 12 named in the evidence - **Tool definitions read at:** [github.com/Alation/alation-ai-agent-sdk/tree/main/guides](https://github.com/Alation/alation-ai-agent-sdk/tree/main/guides) - **Context weight:** 69 tokens for 4 tool definitions (3 with a description), measured from the docs page read - **Status:** Exists on 10 Nov 2026; alation-ai-agent-mcp 1.0.0 on PyPI and GitHub release v1.0.0 dated 26 Oct 2026. Access is read-only by documentation of removed write tools, not by a published read-only switch. - **Read:** 10 Nov 2026 ## In one sentence Alation's server offers a lineage tool and a catalog search agent, and its remote form hides tools the current user cannot use. ## What Alation AI Agent SDK MCP server exposes 12 tools are named in the evidence read on 10 Nov 2026; the tool-definition source measured for the Context Weight Index lists 4. `alation_context`, `bulk_retrieval`, `data_quality_tool`, `get_custom_fields_definitions`, `get_data_products`, `get_data_sources_tool`, `get_signature_creation_instructions`, `lineage`, `catalog_context_search_agent`, `query_flow_agent`, `sql_query_agent`, `custom_agent` ## Not found in the sources read - Documentation that the lineage tool returns transformation SQL/logic or BI-field/metric nodes; the tool page lists graph nodes filtered by object type only. [Source](https://github.com/Alation/alation-ai-agent-sdk/blob/main/guides/tools/lineage.md). - A list of warehouses, transformation or BI tools in the Alation agent SDK README (coverage follows the Alation catalog). [Source](https://github.com/Alation/alation-ai-agent-sdk/blob/main/README.md). - A freshness or collection-cadence statement in the Alation agent SDK README or lineage guide. [Source](https://github.com/Alation/alation-ai-agent-sdk/blob/main/README.md). ## The evidence on Alation AI Agent SDK MCP server, by topic ### Publisher and releases - Alation-ai-agent-mcp is published from the Alation GitHub organization and PyPI, version 1.0.0, released 26 Oct 2026. (observed in the source; [source](https://pypi.org/pypi/alation-ai-agent-mcp/json), read 10 Nov 2026.) - GitHub releases show v1.0.0 published 26 Oct 2026. (observed in the source; [source](https://github.com/Alation/alation-ai-agent-sdk/releases), read 10 Nov 2026.) - The README calls 1.0.0 the first stable release of the SDK. (Alation's own claim; [source](https://github.com/Alation/alation-ai-agent-sdk/blob/main/README.md), read 10 Nov 2026.) ### Clients, auth and scoping - The SDK and its MCP server let AI agents use Alation catalog metadata for curation, search and discovery, and data analyst agents. (Alation's own claim; [source](https://github.com/Alation/alation-ai-agent-sdk/blob/main/README.md), read 10 Nov 2026.) - The remote MCP server lists tools dynamically, filtering out tools the current user cannot use. (Alation's own claim; [source](https://github.com/Alation/alation-ai-agent-sdk/blob/main/README.md), read 10 Nov 2026.) - Service-account authentication gives the agent the data and features allowed by the role assigned to the service account. (Alation's own claim; [source](https://github.com/Alation/alation-ai-agent-sdk/blob/main/guides/authentication.md), read 10 Nov 2026.) - SQL tools need per-data-product data warehouse credentials configured once. (Alation's own claim; [source](https://github.com/Alation/alation-ai-agent-sdk/blob/main/guides/authentication.md), read 10 Nov 2026.) ### Access and safety - Metadata-writing tools update_catalog_asset_metadata and check_job_status were removed from the SDK (Jan 2026), leaving curation to the REST API. (Alation's own claim; [source](https://github.com/Alation/alation-ai-agent-sdk/blob/main/README.md), read 10 Nov 2026.) - The MCP server supports an explicit enabled-tools list (CLI argument or ALATION_ENABLED_TOOLS) to limit the tools exposed. (Alation's own claim; [source](https://github.com/Alation/alation-ai-agent-sdk/blob/main/README.md), read 10 Nov 2026.) - The remote MCP server does not support dynamic client registration; clients must supply a client_id and client_secret. (Alation's own claim; [source](https://github.com/Alation/alation-ai-agent-sdk/blob/main/README.md), read 10 Nov 2026.) ### Lineage and impact - Alation's lineage tool traverses upstream or downstream from a starting object and supports column-level lineage; it is a BETA feature that Alation support must enable on the instance. (Alation's own claim; [source](https://github.com/Alation/alation-ai-agent-sdk/blob/main/guides/tools/lineage.md), read 10 Nov 2026.) - The lineage tool takes max_depth (default 10), allowed_otypes, design/run-time filters and schema exclusions. (Alation's own claim; [source](https://github.com/Alation/alation-ai-agent-sdk/blob/main/guides/tools/lineage.md), read 10 Nov 2026.) ### Definitions, ownership and rules - The lineage docs show combining lineage with the Aggregated Context tool to return stewards or an Owner custom field, and the data dictionary tool for column fields. (Alation's own claim; [source](https://github.com/Alation/alation-ai-agent-sdk/blob/main/guides/tools/lineage.md), read 10 Nov 2026.) - Get_custom_fields_definitions and the data dictionary tools are listed as supported tools for custom metadata. (Alation's own claim; [source](https://github.com/Alation/alation-ai-agent-sdk/blob/main/README.md), read 10 Nov 2026.) - The lineage guide describes updating a PII flag across downstream columns found via lineage, done by the LLM through tools (not automatic propagation). (Alation's own claim; [source](https://github.com/Alation/alation-ai-agent-sdk/blob/main/guides/tools/lineage.md), read 10 Nov 2026.) ### Profile - The remote MCP server is served from the customer's own Alation instance at /ai/mcp. (Alation's own claim; [source](https://github.com/Alation/alation-ai-agent-sdk/blob/main/README.md), read 10 Nov 2026.) ## Read next - [MCP servers for Alation](/mcp/alation/): 1 listed - [Alation AI Agent SDK MCP server and Collibra MCP Server (chip), side by side](/compare/alation-mcp-vs-collibra-mcp/) - [The Context Weight Index](/context-weight/) ## Sources - github.com/Alation/alation-ai-agent-sdk/blob/main/guides/tools/lineage.md: https://github.com/Alation/alation-ai-agent-sdk/blob/main/guides/tools/lineage.md (read 10 Nov 2026) - github.com/Alation/alation-ai-agent-sdk/blob/main/README.md: https://github.com/Alation/alation-ai-agent-sdk/blob/main/README.md (read 10 Nov 2026) - github.com/Alation/alation-ai-agent-sdk/blob/main/guides/authentication.md: https://github.com/Alation/alation-ai-agent-sdk/blob/main/guides/authentication.md (read 10 Nov 2026) - pypi.org/pypi/alation-ai-agent-mcp/json: https://pypi.org/pypi/alation-ai-agent-mcp/json (read 10 Nov 2026) - github.com/Alation/alation-ai-agent-sdk/releases: https://github.com/Alation/alation-ai-agent-sdk/releases (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/servers/alation-mcp/ --- # Amazon Redshift MCP Server (AWS Labs): 8 tools, local (stdio), from AWS Labs SERVER, read 10 Nov 2026 **The short answer.** Amazon Redshift MCP Server (AWS Labs) is an official Labs-org MCP server for Amazon Redshift, published by AWS Labs and read from its own source on 10 Nov 2026. It runs local (stdio), and 8 tools are named. Version read: 0.1.3. Last release: 9 Nov 2026. - **Data tool:** [Amazon Redshift](/mcp/redshift/) - **Publisher:** AWS Labs - **Kind:** Official, Labs org - **Repository or docs:** [github.com/awslabs/mcp/tree/main/src/redshift-mcp-server](https://github.com/awslabs/mcp/tree/main/src/redshift-mcp-server) - **Version or commit:** 0.1.3 - **Last release:** 9 Nov 2026 - **Transport:** Local (stdio) - **Auth:** AWS credentials (profile/env) with IAM and Redshift Data API; database user privileges. - **Access:** Configurable - **Tools listed:** 8 named in the evidence - **Tool definitions read at:** [raw.githubusercontent.com/awslabs/mcp/main/src/redshift-mcp-server/README.md](https://raw.githubusercontent.com/awslabs/mcp/main/src/redshift-mcp-server/README.md) - **Context weight:** 238 tokens for 8 tool definitions (8 with a description), measured from the README read - **Status:** Exists in awslabs/mcp; PyPI 0.1.3 uploaded 9 Nov 2026 (read 10 Nov 2026). - **Read:** 10 Nov 2026 ## In one sentence The AWS Labs server is read-only by default with an explicit read-write opt-in, confirmation prompts and metadata discovery over the Redshift Data API. ## What Amazon Redshift MCP Server (AWS Labs) exposes 8 tools are named in the evidence read on 10 Nov 2026. `list_clusters`, `list_databases`, `list_schemas`, `list_tables`, `list_columns`, `execute_query`, `review_cluster`, `retrieve_skill_file` ## Not found in the sources read - A lineage tool in the Redshift MCP server tool list (eight tools). [Source](https://raw.githubusercontent.com/awslabs/mcp/main/src/redshift-mcp-server/README.md). ## The evidence on Amazon Redshift MCP Server (AWS Labs), by topic ### Publisher and releases - The server lives in the awslabs/mcp repository and PyPI lists author Amazon Web Services; awslabs.redshift-mcp-server 0.1.3 was uploaded 9 Nov 2026. (observed in the source; [source](https://pypi.org/pypi/awslabs.redshift-mcp-server/json), read 10 Nov 2026.) - README calls it a server to discover, explore and query Redshift clusters and serverless workgroups. (AWS Labs's own claim; [source](https://raw.githubusercontent.com/awslabs/mcp/main/src/redshift-mcp-server/README.md), read 10 Nov 2026.) ### Clients, auth and scoping - The README lists 8 tools for discovery, query and cluster review, scoped to the AWS credentials in use. (AWS Labs's own claim; [source](https://raw.githubusercontent.com/awslabs/mcp/main/src/redshift-mcp-server/README.md), read 10 Nov 2026.) ### Access and safety - The server is read-only by default; read-write mode must be selected with ACCESS_MODE. (AWS Labs's own claim; [source](https://raw.githubusercontent.com/awslabs/mcp/main/src/redshift-mcp-server/README.md), read 10 Nov 2026.) - The database user's privileges are the security boundary, not the server. (AWS Labs's own claim; [source](https://raw.githubusercontent.com/awslabs/mcp/main/src/redshift-mcp-server/README.md), read 10 Nov 2026.) - In read-write mode each write prompts for confirmation via MCP elicitation unless UNSAFE_SKIP_WRITE_CONFIRMATION is set. (AWS Labs's own claim; [source](https://raw.githubusercontent.com/awslabs/mcp/main/src/redshift-mcp-server/README.md), read 10 Nov 2026.) - Authentication uses AWS credentials via AWS CLI, profile or environment variables, with IAM permissions for the Data API. (AWS Labs's own claim; [source](https://raw.githubusercontent.com/awslabs/mcp/main/src/redshift-mcp-server/README.md), read 10 Nov 2026.) ### What it reaches - Covers Amazon Redshift provisioned clusters and serverless workgroups. (AWS Labs's own claim; [source](https://raw.githubusercontent.com/awslabs/mcp/main/src/redshift-mcp-server/README.md), read 10 Nov 2026.) ### Lineage and impact - Metadata tools list clusters, databases, schemas, tables and columns; review_cluster reads system views; no lineage tool. (AWS Labs's own claim; [source](https://raw.githubusercontent.com/awslabs/mcp/main/src/redshift-mcp-server/README.md), read 10 Nov 2026.) ### Definitions, ownership and rules - No glossary, ownership or classification tool is listed; retrieve_skill_file returns a Redshift guide covering metadata and SQL. (AWS Labs's own claim; [source](https://raw.githubusercontent.com/awslabs/mcp/main/src/redshift-mcp-server/README.md), read 10 Nov 2026.) ### Freshness and upkeep - Metadata is queried live from the clusters; no collection described. (AWS Labs's own claim; [source](https://raw.githubusercontent.com/awslabs/mcp/main/src/redshift-mcp-server/README.md), read 10 Nov 2026.) ### Profile - Runs over stdio as shown in the README client config. (AWS Labs's own claim; [source](https://raw.githubusercontent.com/awslabs/mcp/main/src/redshift-mcp-server/README.md), read 10 Nov 2026.) ## Read next - [MCP servers for Amazon Redshift](/mcp/redshift/): 1 listed - [The Context Weight Index](/context-weight/) ## Sources - pypi.org/pypi/awslabs.redshift-mcp-server/json: https://pypi.org/pypi/awslabs.redshift-mcp-server/json (read 10 Nov 2026) - raw.githubusercontent.com/awslabs/mcp/main/src/redshift-mcp-server/README.md: https://raw.githubusercontent.com/awslabs/mcp/main/src/redshift-mcp-server/README.md (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/servers/awslabs-redshift-mcp-server/ --- # Atlan MCP Server: 41 tools, remote (hosted), from Atlan SERVER, read 10 Nov 2026 **The short answer.** Atlan MCP Server is an official MCP server for Atlan, published by Atlan and read from its own source on 10 Nov 2026. It runs remote (hosted), and 41 tools are named. Version read: Hosted at mcp.atlan.com/mcp (unversioned); local atlan-mcp-server 0.3.3 (deprecated). Last release: 17 Feb 2026. - **Data tool:** [Atlan](/mcp/atlan/) - **Publisher:** Atlan - **Kind:** Official - **Repository or docs:** [docs.atlan.com/product/capabilities/atlan-ai/references/mcp-tools](https://docs.atlan.com/product/capabilities/atlan-ai/references/mcp-tools) - **Version or commit:** Hosted at mcp.atlan.com/mcp (unversioned); local atlan-mcp-server 0.3.3 (deprecated) - **Last release:** 17 Feb 2026 - **Transport:** Remote (hosted) - **Auth:** OAuth per user (calls run as the signed-in user) or Atlan API key as Bearer token (shared service identity); local server uses API key + base URL. - **Access:** Configurable - **Tools listed:** 41 named in the evidence - **Tool definitions read at:** [docs.atlan.com/product/capabilities/atlan-ai/references/mcp-tools](https://docs.atlan.com/product/capabilities/atlan-ai/references/mcp-tools) - **Context weight:** 1,770 tokens for 42 tool definitions (42 with a description), measured from the docs page read - **Status:** Exists on 10 Nov 2026: hosted Atlan MCP (41 tools, page updated 27 Oct 2026); the local atlan-mcp-server in atlanhq/agent-toolkit is marked deprecated in its README. - **Read:** 10 Nov 2026 ## In one sentence Atlan's hosted per-tenant server runs as the signed-in user, scopes context by persona, offers an optional read-only mode and proposes writes for approval. ## What Atlan MCP Server exposes 41 tools are named in the evidence read on 10 Nov 2026; the tool-definition source measured for the Context Weight Index lists 42. `semantic_search`, `ask`, `fluent_search`, `fluent_search_schema`, `search_assets`, `get_assets`, `search_atlan_docs`, `resolve_metadata`, `get_groups`, `get_users`, `describe_asset_type`, `traverse_lineage`, `query_assets`, `update_assets`, `update_custom_metadata`, `add_atlan_tags`, `remove_atlan_tag`, `manage_announcements`, `create_custom_metadata_set`, `add_attributes_to_cm_set`, `remove_attributes_from_cm_set`, `remove_custom_metadata`, `delete_custom_metadata_set`, `manage_asset_lifecycle`, `create_glossaries`, `create_glossary_categories`, `create_glossary_terms`, `create_domains`, `create_data_products`, `create_dq_rules`, `update_dq_rules`, `schedule_dq_rules`, `delete_dq_rules`, `read_artifact`, `search_knowledge`, `list_knowledge`, `read_knowledge_file`, `create_knowledge_folder`, `save_knowledge_content`, `get_knowledge_upload_url`, `register_knowledge_file` ## Not found in the sources read - A statement on the MCP tools page that traverse_lineage returns column-level edges or transformation SQL. [Source](https://docs.atlan.com/product/capabilities/atlan-ai/references/mcp-tools). ## The evidence on Atlan MCP Server, by topic ### Publisher and releases - Atlan hosts the MCP server itself at mcp.atlan.com/mcp, per tenant. (Atlan's own claim; [source](https://docs.atlan.com/product/capabilities/atlan-ai/references/mcp-security), read 10 Nov 2026.) - The older local server in atlanhq/agent-toolkit is deprecated in favor of the hosted endpoint. (Atlan's own claim; [source](https://github.com/atlanhq/agent-toolkit/blob/main/modelcontextprotocol/README.md), read 10 Nov 2026.) - The last GitHub release of the local server is v0.3.3, on 17 Feb 2026; the hosted server has no public version number. (observed in the source; [source](https://github.com/atlanhq/agent-toolkit/blob/main/CHANGELOG.md), read 10 Nov 2026.) ### Clients, auth and scoping - The MCP server serves the catalog as a context layer to AI clients (Claude, Cursor, ChatGPT, Gemini, Copilot) and automation platforms. (Atlan's own claim; [source](https://docs.atlan.com/product/capabilities/atlan-ai/how-tos/remote-mcp-overview), read 10 Nov 2026.) - OAuth makes each tool call run as the signed-in user, so tools run with their own permissions and actions are attributed to them. (Atlan's own claim; [source](https://docs.atlan.com/product/capabilities/atlan-ai/how-tos/remote-mcp-overview), read 10 Nov 2026.) - Guests and members are scoped to their persona; MCP narrows but never widens what a user's login permits. (Atlan's own claim; [source](https://docs.atlan.com/product/capabilities/atlan-ai/references/mcp-security), read 10 Nov 2026.) - The tools page lists 41 tools across nine categories. (Atlan's own claim; [source](https://docs.atlan.com/product/capabilities/atlan-ai/references/mcp-tools), read 10 Nov 2026.) ### Access and safety - An optional read-only mode removes every write, admin and lifecycle tool while keeping search and lineage; it is requested from Atlan support. (Atlan's own claim; [source](https://docs.atlan.com/product/capabilities/atlan-ai/references/mcp-security), read 10 Nov 2026.) - Tools that change anything propose first and wait for approval before writing. (Atlan's own claim; [source](https://docs.atlan.com/product/capabilities/atlan-ai/references/mcp-tools), read 10 Nov 2026.) - Every tool call is logged as a structured event with tool name, client, tenant, request ID and status. (Atlan's own claim; [source](https://docs.atlan.com/product/capabilities/atlan-ai/references/mcp-security), read 10 Nov 2026.) - Tools that execute SQL route it through a read-only validator that permits only SELECT-class queries. (Atlan's own claim; [source](https://docs.atlan.com/product/capabilities/atlan-ai/references/mcp-security), read 10 Nov 2026.) ### What it reaches - The docs list supported AI clients: Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, ChatGPT, Snowflake, Google ADK, n8n, Microsoft 365 Copilot, Glean and Databricks. Data-source coverage depends on Atlan connectors, not stated on this page. (Atlan's own claim; [source](https://docs.atlan.com/product/capabilities/atlan-ai/how-tos/remote-mcp-overview), read 10 Nov 2026.) ### Lineage and impact - The hosted Atlan MCP exposes a traverse_lineage tool for upstream/downstream lineage with deeper hops on request. (Atlan's own claim; [source](https://docs.atlan.com/product/capabilities/atlan-ai/references/mcp-tools), read 10 Nov 2026.) - The docs say the hosted Atlan MCP can traverse end-to-end lineage; the tool page does not state column-level or SQL-logic parsing for the lineage tool. (Atlan's own claim; [source](https://docs.atlan.com/product/capabilities/atlan-ai/how-tos/remote-mcp-overview), read 10 Nov 2026.) - Query_assets runs read-only SQL against a connected warehouse (Snowflake, Redshift, BigQuery, Databricks), returning up to 100 rows. (Atlan's own claim; [source](https://docs.atlan.com/product/capabilities/atlan-ai/references/mcp-tools), read 10 Nov 2026.) ### Definitions, ownership and rules - Tools cover glossaries, glossary categories and terms, domains and data products, custom metadata sets, and certification/descriptions. (Atlan's own claim; [source](https://docs.atlan.com/product/capabilities/atlan-ai/references/mcp-tools), read 10 Nov 2026.) - Add_atlan_tags applies classification tags with propagation to downstream lineage on by default. (Atlan's own claim; [source](https://docs.atlan.com/product/capabilities/atlan-ai/references/mcp-tools), read 10 Nov 2026.) - Resolve_metadata returns tenant-level tag and custom-metadata definitions reconciled live, including enum option values. (Atlan's own claim; [source](https://docs.atlan.com/product/capabilities/atlan-ai/references/mcp-tools), read 10 Nov 2026.) - The lineage tool returns owners, certificate status, tags and readme with each asset by default (local server docstring). (observed in the source; [source](https://github.com/atlanhq/agent-toolkit/blob/main/modelcontextprotocol/server.py), read 10 Nov 2026.) ### Freshness and upkeep - Metadata is retrieved in real time with no intermediate caching. (Atlan's own claim; [source](https://docs.atlan.com/product/capabilities/atlan-ai/references/mcp-security), read 10 Nov 2026.) ### Profile - Atlan's compliance posture for MCP is covered by HIPAA, GDPR and CCPA/CPRA programs, with certifications on the Atlan Trust Portal. (Atlan's own claim; [source](https://docs.atlan.com/product/capabilities/atlan-ai/references/mcp-security), read 10 Nov 2026.) ## Read next - [MCP servers for Atlan](/mcp/atlan/): 1 listed - [Atlan MCP Server and DataHub MCP Server, side by side](/compare/atlan-mcp-vs-datahub-mcp/) - [Atlan MCP Server and Euno MCP, side by side](/compare/atlan-mcp-vs-euno-mcp/) - [The Context Weight Index](/context-weight/) ## Sources - docs.atlan.com/product/capabilities/atlan-ai/references/mcp-tools: https://docs.atlan.com/product/capabilities/atlan-ai/references/mcp-tools (read 10 Nov 2026) - docs.atlan.com/product/capabilities/atlan-ai/how-tos/remote-mcp-overview: https://docs.atlan.com/product/capabilities/atlan-ai/how-tos/remote-mcp-overview (read 10 Nov 2026) - github.com/atlanhq/agent-toolkit/blob/main/modelcontextprotocol/server.py: https://github.com/atlanhq/agent-toolkit/blob/main/modelcontextprotocol/server.py (read 10 Nov 2026) - docs.atlan.com/product/capabilities/atlan-ai/references/mcp-security: https://docs.atlan.com/product/capabilities/atlan-ai/references/mcp-security (read 10 Nov 2026) - github.com/atlanhq/agent-toolkit/blob/main/modelcontextprotocol/README.md: https://github.com/atlanhq/agent-toolkit/blob/main/modelcontextprotocol/README.md (read 10 Nov 2026) - github.com/atlanhq/agent-toolkit/blob/main/CHANGELOG.md: https://github.com/atlanhq/agent-toolkit/blob/main/CHANGELOG.md (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/servers/atlan-mcp/ --- # BigQuery remote MCP server: 9 tools, remote (hosted), from Google Cloud SERVER, read 10 Nov 2026 **The short answer.** BigQuery remote MCP server is an official MCP server for BigQuery, published by Google Cloud and read from its own source on 10 Nov 2026. It runs remote (hosted), and 9 tools are named. Version read: Docs reference last updated 3 Nov 2026. Last release: none found on 10 Nov 2026. - **Data tool:** [BigQuery](/mcp/bigquery/) - **Publisher:** Google Cloud - **Kind:** Official - **Repository or docs:** [docs.cloud.google.com/bigquery/docs/use-bigquery-mcp](https://docs.cloud.google.com/bigquery/docs/use-bigquery-mcp) - **Version or commit:** Docs reference last updated 3 Nov 2026 - **Last release:** none found on 10 Nov 2026 - **Transport:** Remote (hosted) - **Auth:** OAuth 2.0 with IAM (Google Cloud identities, OAuth client, or agent identity). - **Access:** Configurable - **Tools listed:** 9 named in the evidence - **Tool definitions read at:** [docs.cloud.google.com/bigquery/docs/reference/mcp](https://docs.cloud.google.com/bigquery/docs/reference/mcp) - **Context weight:** 1,487 tokens for 9 tool definitions (9 with a description), measured from the docs page read - **Status:** Exists; Google-managed endpoint, reference page last updated 3 Nov 2026 (read 10 Nov 2026). - **Read:** 10 Nov 2026 ## In one sentence Google Cloud's managed endpoint gives IAM-scoped access to a SELECT-only query tool and dataset and table metadata tools. ## What BigQuery remote MCP server exposes 9 tools are named in the evidence read on 10 Nov 2026. `list_dataset_ids`, `get_dataset_info`, `list_table_ids`, `get_table_info`, `execute_sql_readonly`, `execute_sql`, `get_query_results`, `cancel_job`, `get_job` ## Not found in the sources read - A lineage or Dataplex lineage tool in the BigQuery MCP tools reference (nine tools listed). [Source](https://docs.cloud.google.com/bigquery/docs/reference/mcp). ## The evidence on BigQuery remote MCP server, by topic ### Publisher and releases - Google documents a BigQuery remote MCP server, enabled when the BigQuery API is enabled. (Google Cloud's own claim; [source](https://docs.cloud.google.com/bigquery/docs/use-bigquery-mcp), read 10 Nov 2026.) - The reference page for bigquery.googleapis.com MCP shows Last updated 3 Nov 2026 UTC; no semantic version is published. (Google Cloud's own claim; [source](https://docs.cloud.google.com/bigquery/docs/reference/mcp), read 10 Nov 2026.) ### Clients, auth and scoping - The server can run queries, get metadata and list resources for AI apps including Claude and ChatGPT. (Google Cloud's own claim; [source](https://docs.cloud.google.com/bigquery/docs/use-bigquery-mcp), read 10 Nov 2026.) - OAuth 2.0 with IAM is used so access follows the authenticated Google Cloud identity. (Google Cloud's own claim; [source](https://docs.cloud.google.com/bigquery/docs/use-bigquery-mcp), read 10 Nov 2026.) ### Access and safety - Execute_sql_readonly is restricted to SELECT statements and is preferred over execute_sql. (Google Cloud's own claim; [source](https://docs.cloud.google.com/bigquery/docs/reference/mcp), read 10 Nov 2026.) - Execute_sql can run INSERT, UPDATE, DELETE and CREATE statements. (Google Cloud's own claim; [source](https://docs.cloud.google.com/bigquery/docs/reference/mcp), read 10 Nov 2026.) - Google remote MCP servers offer centralized audit logging and optional Model Armor protection. (Google Cloud's own claim; [source](https://docs.cloud.google.com/bigquery/docs/use-bigquery-mcp), read 10 Nov 2026.) ### What it reaches - Covers BigQuery and BigLake tables and namespaces. (Google Cloud's own claim; [source](https://docs.cloud.google.com/bigquery/docs/reference/mcp), read 10 Nov 2026.) ### Lineage and impact - Tools cover dataset and table metadata and SQL execution; no lineage tool among the nine listed. (Google Cloud's own claim; [source](https://docs.cloud.google.com/bigquery/docs/reference/mcp), read 10 Nov 2026.) ### Definitions, ownership and rules - Get_table_info and get_dataset_info return metadata for BigQuery and BigLake objects; no glossary or policy tool is listed. (Google Cloud's own claim; [source](https://docs.cloud.google.com/bigquery/docs/reference/mcp), read 10 Nov 2026.) ### Freshness and upkeep - Calls hit the live BigQuery APIs; the server has no quotas of its own beyond the underlying API quotas. (Google Cloud's own claim; [source](https://docs.cloud.google.com/bigquery/docs/use-bigquery-mcp), read 10 Nov 2026.) ### Profile - Global endpoint bigquery.googleapis.com/mcp with Streamable HTTP transport. (Google Cloud's own claim; [source](https://docs.cloud.google.com/bigquery/docs/use-bigquery-mcp), read 10 Nov 2026.) ## Read next - [MCP servers for BigQuery](/mcp/bigquery/): 2 listed - [MCP Toolbox for Databases (BigQuery tools)](/servers/mcp-toolbox-bigquery/): also for BigQuery - [BigQuery remote MCP server and MCP Toolbox for Databases (BigQuery tools), side by side](/compare/bigquery-remote-mcp-vs-mcp-toolbox-bigquery/) - [The Context Weight Index](/context-weight/) ## Sources - docs.cloud.google.com/bigquery/docs/use-bigquery-mcp: https://docs.cloud.google.com/bigquery/docs/use-bigquery-mcp (read 10 Nov 2026) - docs.cloud.google.com/bigquery/docs/reference/mcp: https://docs.cloud.google.com/bigquery/docs/reference/mcp (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/servers/bigquery-remote-mcp/ --- # MCP Toolbox for Databases (BigQuery tools): 9 tools, remote and local, from Google SERVER, read 10 Nov 2026 **The short answer.** MCP Toolbox for Databases (BigQuery tools) is an official MCP server for BigQuery, published by Google and read from its own source on 10 Nov 2026. It runs remote and local, and 9 tools are named. Version read: V1.16.0. Last release: 5 Nov 2026. - **Data tool:** [BigQuery](/mcp/bigquery/) - **Publisher:** Google (googleapis) - **Kind:** Official - **Repository or docs:** [github.com/googleapis/mcp-toolbox](https://github.com/googleapis/mcp-toolbox) - **Version or commit:** V1.16.0 - **Last release:** 5 Nov 2026 - **Transport:** Remote and local - **Auth:** Application Default Credentials, client OAuth token forwarding, or service account impersonation. - **Access:** Configurable - **Tools listed:** 9 named in the evidence - **Tool definitions read at:** [raw.githubusercontent.com/googleapis/mcp-toolbox/main/docs/en/integrations/bigquery/prebuilt-configs/bigquery.md](https://raw.githubusercontent.com/googleapis/mcp-toolbox/main/docs/en/integrations/bigquery/prebuilt-configs/bigquery.md) - **Context weight:** 156 tokens for 9 tool definitions (9 with a description), measured from the docs page read - **Status:** Repo googleapis/genai-toolbox now lives at googleapis/mcp-toolbox (read 10 Nov 2026); BigQuery tools documented there. - **Read:** 10 Nov 2026 ## In one sentence Google's open-source Toolbox offers BigQuery SQL and metadata tools with a read-only mode, dataset allowlists and Knowledge Catalog search. ## What MCP Toolbox for Databases (BigQuery tools) exposes 9 tools are named in the evidence read on 10 Nov 2026. `analyze_contribution`, `ask_data_insights`, `execute_sql`, `forecast`, `get_dataset_info`, `get_table_info`, `list_dataset_ids`, `list_table_ids`, `search_catalog` ## Not found in the sources read - A lineage tool in the BigQuery prebuilt tool list. [Source](https://raw.githubusercontent.com/googleapis/mcp-toolbox/main/docs/en/integrations/bigquery/prebuilt-configs/bigquery.md). ## The evidence on MCP Toolbox for Databases (BigQuery tools), by topic ### Publisher and releases - MCP Toolbox for Databases is an open source MCP server in the googleapis organization (formerly genai-toolbox) with a BigQuery prebuilt config. (Google's own claim; [source](https://raw.githubusercontent.com/googleapis/mcp-toolbox/main/README.md), read 10 Nov 2026.) - The latest GitHub release is v1.16.0, whose release timestamp on the page is 5 Nov 2026. (observed in the source; [source](https://github.com/googleapis/mcp-toolbox/releases), read 10 Nov 2026.) ### Clients, auth and scoping - Tools include ask_data_insights (Conversational Analytics API), forecast and analyze_contribution. (Google's own claim; [source](https://raw.githubusercontent.com/googleapis/mcp-toolbox/main/docs/en/integrations/bigquery/prebuilt-configs/bigquery.md), read 10 Nov 2026.) - With useClientOAuth the client's OAuth token is forwarded so calls run as the asking user. (Google's own claim; [source](https://raw.githubusercontent.com/googleapis/mcp-toolbox/main/docs/en/integrations/bigquery/source.md), read 10 Nov 2026.) ### Access and safety - Source options readOnly and writeMode (allowed, blocked, protected) enforce read-only operation; blocked allows only SELECT. (Google's own claim; [source](https://raw.githubusercontent.com/googleapis/mcp-toolbox/main/docs/en/integrations/bigquery/source.md), read 10 Nov 2026.) - AllowedDatasets restricts tools to listed datasets and impersonateServiceAccount supports service account impersonation. (Google's own claim; [source](https://raw.githubusercontent.com/googleapis/mcp-toolbox/main/docs/en/integrations/bigquery/source.md), read 10 Nov 2026.) - BIGQUERY_MAXIMUM_BYTES_BILLED caps per-query bytes scanned. (Google's own claim; [source](https://raw.githubusercontent.com/googleapis/mcp-toolbox/main/docs/en/integrations/bigquery/prebuilt-configs/bigquery.md), read 10 Nov 2026.) ### What it reaches - The BigQuery source covers BigQuery; the Toolbox itself supports other databases. (Google's own claim; [source](https://raw.githubusercontent.com/googleapis/mcp-toolbox/main/README.md), read 10 Nov 2026.) ### Lineage and impact - The BigQuery prebuilt includes search_catalog which searches Knowledge Catalog (formerly Dataplex) entries; no lineage tool is listed. (Google's own claim; [source](https://raw.githubusercontent.com/googleapis/mcp-toolbox/main/docs/en/integrations/bigquery/tools/bigquery-search-catalog.md), read 10 Nov 2026.) ### Definitions, ownership and rules - Search_catalog finds tables, views, models, routines and connections via Knowledge Catalog metadata. (Google's own claim; [source](https://raw.githubusercontent.com/googleapis/mcp-toolbox/main/docs/en/integrations/bigquery/tools/bigquery-search-catalog.md), read 10 Nov 2026.) ### Freshness and upkeep - Tools query BigQuery and Knowledge Catalog live through ADC or OAuth; no collection mechanism. (Google's own claim; [source](https://raw.githubusercontent.com/googleapis/mcp-toolbox/main/docs/en/integrations/bigquery/tools/bigquery-search-catalog.md), read 10 Nov 2026.) ### Profile - The README shows running with a --stdio flag for MCP clients. (Google's own claim; [source](https://raw.githubusercontent.com/googleapis/mcp-toolbox/main/README.md), read 10 Nov 2026.) ## Read next - [MCP servers for BigQuery](/mcp/bigquery/): 2 listed - [BigQuery remote MCP server](/servers/bigquery-remote-mcp/): also for BigQuery - [BigQuery remote MCP server and MCP Toolbox for Databases (BigQuery tools), side by side](/compare/bigquery-remote-mcp-vs-mcp-toolbox-bigquery/) - [The Context Weight Index](/context-weight/) ## Sources - raw.githubusercontent.com/googleapis/mcp-toolbox/main/README.md: https://raw.githubusercontent.com/googleapis/mcp-toolbox/main/README.md (read 10 Nov 2026) - github.com/googleapis/mcp-toolbox/releases: https://github.com/googleapis/mcp-toolbox/releases (read 10 Nov 2026) - raw.githubusercontent.com/googleapis/mcp-toolbox/main/docs/en/integrations/bigquery/tools/bigquery-search-catalog.md: https://raw.githubusercontent.com/googleapis/mcp-toolbox/main/docs/en/integrations/bigquery/tools/bigquery-search-catalog.md (read 10 Nov 2026) - raw.githubusercontent.com/googleapis/mcp-toolbox/main/docs/en/integrations/bigquery/prebuilt-configs/bigquery.md: https://raw.githubusercontent.com/googleapis/mcp-toolbox/main/docs/en/integrations/bigquery/prebuilt-configs/bigquery.md (read 10 Nov 2026) - raw.githubusercontent.com/googleapis/mcp-toolbox/main/docs/en/integrations/bigquery/source.md: https://raw.githubusercontent.com/googleapis/mcp-toolbox/main/docs/en/integrations/bigquery/source.md (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/servers/mcp-toolbox-bigquery/ --- # Collibra MCP Server (chip): 49 tools, remote and local, from Collibra SERVER, read 10 Nov 2026 **The short answer.** Collibra MCP Server (chip) is an official MCP server for Collibra, published by Collibra and read from its own source on 10 Nov 2026. It runs remote and local, and 49 tools are named. Version read: V0.0.67. Last release: 27 Oct 2026. 23 sourced facts follow. - **Data tool:** [Collibra](/mcp/collibra/) - **Publisher:** Collibra - **Kind:** Official - **Repository or docs:** [github.com/collibra/chip](https://github.com/collibra/chip) - **Version or commit:** V0.0.67 - **Last release:** 27 Oct 2026 - **Transport:** Remote and local - **Auth:** Collibra username/password: server-wide credentials over stdio (mcp.yaml or env vars), or client-supplied Basic Auth headers per request over HTTP; tools respect the connecting user's DGC scopes. - **Access:** Configurable - **Tools listed:** 49 named in the evidence - **Tool definitions read at:** [github.com/collibra/chip/blob/main/README.md](https://github.com/collibra/chip/blob/main/README.md) - **Context weight:** 2,914 tokens for 41 tool definitions (41 with a description), measured from the README read - **Status:** Exists on 10 Nov 2026 as collibra/chip; v0.0.67 released 27 Oct 2026 per GitHub releases. Collibra's product-docs page for it returned a Cloudflare block, so only the repo README was read. - **Read:** 10 Nov 2026 ## In one sentence Collibra's server covers lineage with transformation logic, business terms and column semantics, classification, stewardship edits and data quality, under the connecting user's permissions. ## What Collibra MCP Server (chip) exposes 49 tools are named in the evidence read on 10 Nov 2026; the tool-definition source measured for the Context Weight Index lists 41. `discover_business_glossary`, `discover_data_assets`, `get_assessment`, `get_asset_details`, `get_business_term_data`, `get_column_semantics`, `get_data_quality_rule`, `get_data_quality_rule_results`, `validate_data_quality_rule`, `list_data_quality_rule_templates`, `get_data_quality_rule_template`, `find_data_quality_rules`, `generate_data_quality_rule_sql`, `get_lineage_downstream`, `get_lineage_entity`, `get_lineage_transformation`, `get_lineage_upstream`, `get_measure_data`, `get_table_semantics`, `get_user_id_by_name`, `list_asset_types`, `list_data_contract`, `prepare_create_asset`, `pull_data_contract_manifest`, `search_asset_keyword`, `search_catalog_columns`, `search_data_class`, `search_data_classification_match`, `search_lineage_entities`, `search_lineage_transformations`, `add_data_classification_match`, `create_assessment`, `create_asset`, `create_data_quality_job`, `create_data_quality_rule`, `deploy_data_quality_rule_template`, `dq_cancel_job_run`, `dq_delete_job`, `dq_delete_job_run`, `dq_update_job`, `dq_get_job`, `dq_get_job_run`, `dq_search_jobs`, `dq_search_job_runs`, `edit_assessment`, `edit_asset`, `init_data_contract`, `push_data_contract_manifest`, `remove_data_classification_match` ## Not found in the sources read - A README statement that the lineage tools return column-level edges or BI-field lineage explicitly; the tool list names entities and transformations only. [Source](https://github.com/collibra/chip/blob/main/README.md). - A list of warehouses, BI tools or transformation tools supported by the Collibra MCP README (coverage depends on what is catalogued in the Collibra instance). [Source](https://github.com/collibra/chip/blob/main/README.md). - Any statement on collection cadence or freshness in the Collibra MCP README (the server queries a live Collibra instance). [Source](https://github.com/collibra/chip/blob/main/README.md). ## The evidence on Collibra MCP Server (chip), by topic ### Publisher and releases - The repository lives under the collibra GitHub organization and is titled the Collibra MCP Server (Apache-2.0). (Collibra's own claim; [source](https://github.com/collibra/chip/blob/main/README.md), read 10 Nov 2026.) - GitHub releases show v0.0.67 published 27 Oct 2026, so the server is still at a 0.0.x version. (observed in the source; [source](https://github.com/collibra/chip/releases), read 10 Nov 2026.) - The release timestamp for v0.0.67 is 27 Oct 2026. (observed in the source; [source](https://github.com/collibra/chip/releases), read 10 Nov 2026.) ### Clients, auth and scoping - The server is described as giving AI agents access to Collibra Data Governance Center for asset discovery, glossary queries and asset details. (Collibra's own claim; [source](https://github.com/collibra/chip/blob/main/README.md), read 10 Nov 2026.) - Permission scoping follows the connecting Collibra user: tools needing extra DGC scopes are flagged, and a missing scope returns a permission error. (Collibra's own claim; [source](https://github.com/collibra/chip/blob/main/README.md), read 10 Nov 2026.) - Natural-language discovery tools use Collibra AI and consume Collibra Units. (Collibra's own claim; [source](https://github.com/collibra/chip/blob/main/README.md), read 10 Nov 2026.) - With the HTTP transport, MCP clients are advised to supply their own Basic Auth headers per request; with stdio, one server-wide credential is used. (Collibra's own claim; [source](https://github.com/collibra/chip/blob/main/README.md), read 10 Nov 2026.) ### Access and safety - Data quality tools (18 tools, several that write or delete) are off by default and gated behind a --data-quality flag. (Collibra's own claim; [source](https://github.com/collibra/chip/blob/main/README.md), read 10 Nov 2026.) - Specific tools can be enabled or disabled through flags, environment variables or mcp.yaml; by default all are enabled. (Collibra's own claim; [source](https://github.com/collibra/chip/blob/main/README.md), read 10 Nov 2026.) - Destructive DQ tools use a confirm checkpoint (confirm=false by default is read-only). (Collibra's own claim; [source](https://github.com/collibra/chip/blob/main/README.md), read 10 Nov 2026.) - HTTP mode binds to localhost only, and credentials should be kept in environment variables. (Collibra's own claim; [source](https://github.com/collibra/chip/blob/main/README.md), read 10 Nov 2026.) ### Lineage and impact - The Collibra MCP server has separate tools for upstream and downstream technical lineage, lineage entities, and transformations. (Collibra's own claim; [source](https://github.com/collibra/chip/blob/main/README.md), read 10 Nov 2026.) - Get_lineage_transformation returns the details and logic of a specific data transformation. (Collibra's own claim; [source](https://github.com/collibra/chip/blob/main/README.md), read 10 Nov 2026.) - Lineage tools search entities and transformations in the technical lineage graph. (Collibra's own claim; [source](https://github.com/collibra/chip/blob/main/README.md), read 10 Nov 2026.) - Get_measure_data traces a measure back to its underlying physical columns and tables. (Collibra's own claim; [source](https://github.com/collibra/chip/blob/main/README.md), read 10 Nov 2026.) ### Definitions, ownership and rules - Get_business_term_data traces a business term back to its connected physical data assets. (Collibra's own claim; [source](https://github.com/collibra/chip/blob/main/README.md), read 10 Nov 2026.) - Get_column_semantics returns data attributes, measures and business assets connected to a column; get_table_semantics returns the table's semantic layer. (Collibra's own claim; [source](https://github.com/collibra/chip/blob/main/README.md), read 10 Nov 2026.) - Edit_asset can set attributes such as Definition and assign users or groups to resource roles such as Steward and Owner. (Collibra's own claim; [source](https://github.com/collibra/chip/blob/main/README.md), read 10 Nov 2026.) - Classification tools associate data classes with assets (add_data_classification_match, search_data_classification_match). (Collibra's own claim; [source](https://github.com/collibra/chip/blob/main/README.md), read 10 Nov 2026.) ### Profile - The server is a Go binary distributed as prebuilt releases for Linux, macOS and Windows. (Collibra's own claim; [source](https://github.com/collibra/chip/blob/main/README.md), read 10 Nov 2026.) ## Read next - [MCP servers for Collibra](/mcp/collibra/): 1 listed - [Alation AI Agent SDK MCP server and Collibra MCP Server (chip), side by side](/compare/alation-mcp-vs-collibra-mcp/) - [Collibra MCP Server (chip) and Euno MCP, side by side](/compare/collibra-mcp-vs-euno-mcp/) - [The Context Weight Index](/context-weight/) ## Sources - github.com/collibra/chip/blob/main/README.md: https://github.com/collibra/chip/blob/main/README.md (read 10 Nov 2026) - github.com/collibra/chip/releases: https://github.com/collibra/chip/releases (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/servers/collibra-mcp/ --- # Cube MCP server: 30 tools, remote (hosted), from Cube SERVER, read 10 Nov 2026 **The short answer.** Cube MCP server is an official MCP server for Cube, published by Cube and read from its own source on 10 Nov 2026. It runs remote (hosted), and 30 tools are named. Version read: Hosted service (no versioned release). Last release: none found on 10 Nov 2026. - **Data tool:** [Cube](/mcp/cube/) - **Publisher:** Cube - **Kind:** Official - **Repository or docs:** [docs.cube.dev/docs/integrations/mcp-server](https://docs.cube.dev/docs/integrations/mcp-server) - **Version or commit:** Hosted service (no versioned release) - **Last release:** none found on 10 Nov 2026 - **Transport:** Remote (hosted) - **Auth:** OAuth Authorization Code + PKCE (scope mcp-agent-access, public client cube-mcp-client), or a Cube API key exchanged for a token (client_id cube-api-key). - **Access:** Read and write - **Tools listed:** 30 named in the evidence - **Tool definitions read at:** [docs.cube.dev/docs/integrations/mcp-server.md](https://docs.cube.dev/docs/integrations/mcp-server.md) - **Context weight:** 1,093 tokens for 30 tool definitions (30 with a description), measured from the docs page read - **Status:** Exists on 10 Nov 2026 as a Cube-hosted endpoint cubecloud.dev/mcp (or your own console domain). The page says it exposes 30 tools; the tables list 28 names, all captured here. - **Read:** 10 Nov 2026 ## In one sentence Cube's hosted server searches the semantic layer, runs SQL with row-level security as the signed-in user and edits the semantic model on a branch. ## What Cube MCP server exposes 30 tools are named in the evidence read on 10 Nov 2026. `listDeployments`, `chat`, `loadQueryResults`, `searchDataModel`, `runQuery`, `readWorkbook`, `createWorkbook`, `createReport`, `readReport`, `updateReport`, `deleteReport`, `manageFolders`, `updateDashboard`, `publishDashboard`, `listDataModelFiles`, `readDataModelFile`, `startDataModelEdit`, `writeDataModelFile`, `deleteDataModelFile`, `getDataModelChanges`, `getBranchDiff`, `getDeploymentEnv`, `getBranchState`, `commitDataModelChanges`, `commitToNewBranch`, `mergeToDefaultBranch`, `discardDataModelChanges`, `switchUserBranch`, `getPreAggregationStatus`, `buildPreAggregation` ## Not found in the sources read - A lineage tool in the Cube MCP tool tables (getBranchDiff and getDataModelChanges compare model versions, not data lineage). [Source](https://docs.cube.dev/docs/integrations/mcp-server.md). ## The evidence on Cube MCP server, by topic ### Publisher and releases - Cube hosts the MCP server and documents it on docs.cube.dev, available on all plans. (Cube's own claim; [source](https://docs.cube.dev/docs/integrations/mcp-server.md), read 10 Nov 2026.) - No versioned release is published; one endpoint cubecloud.dev/mcp serves every account, deployment and region. (Cube's own claim; [source](https://docs.cube.dev/docs/integrations/mcp-server.md), read 10 Nov 2026.) ### Clients, auth and scoping - Every tool runs as the authenticated user, and queries respect Cube permissions including row-level security. (Cube's own claim; [source](https://docs.cube.dev/docs/integrations/mcp-server.md), read 10 Nov 2026.) - With an API key, a token can act as the key's owner, as another user, or be limited to specific tools. (Cube's own claim; [source](https://docs.cube.dev/docs/integrations/mcp-server.md), read 10 Nov 2026.) - Which tools a user sees depends on their role; Viewer role or higher is needed, and data model editing tools show only for roles allowed to edit the model (Admin and Developer by default). (Cube's own claim; [source](https://docs.cube.dev/docs/integrations/mcp-server.md), read 10 Nov 2026.) ### Access and safety - Nine tools are annotated destructive and clients that honor annotations always ask before running them. (Cube's own claim; [source](https://docs.cube.dev/docs/integrations/mcp-server.md), read 10 Nov 2026.) - Chat is annotated read-only but runs the Analytics Chat agent, which can create reports or delete folders without confirmation, so the annotation understates what it can do. (Cube's own claim; [source](https://docs.cube.dev/docs/integrations/mcp-server.md), read 10 Nov 2026.) - File writes never touch production: writeDataModelFile and deleteDataModelFile reject any branch that isn't a dev branch. (Cube's own claim; [source](https://docs.cube.dev/docs/integrations/mcp-server.md), read 10 Nov 2026.) - API-key tokens last up to one hour and there is no refresh token. (Cube's own claim; [source](https://docs.cube.dev/docs/integrations/mcp-server.md), read 10 Nov 2026.) ### What it reaches - Documented clients are Claude, ChatGPT, Microsoft 365 Copilot, Cursor and Codex, plus other MCP clients over HTTPS. (Cube's own claim; [source](https://docs.cube.dev/docs/integrations/mcp-server.md), read 10 Nov 2026.) - Admins choose the default deployment and can limit which deployments MCP clients can reach, always intersected with role-based permissions. (Cube's own claim; [source](https://docs.cube.dev/docs/integrations/mcp-server.md), read 10 Nov 2026.) ### Lineage and impact - SearchDataModel searches the semantic model by similarity, covering views and their measures and dimensions, and returns name, title, description and type. (Cube's own claim; [source](https://docs.cube.dev/docs/integrations/mcp-server.md), read 10 Nov 2026.) - ListDataModelFiles and readDataModelFile read the raw YAML, JavaScript or Python source of the semantic model, so the definition logic itself is readable. (Cube's own claim; [source](https://docs.cube.dev/docs/integrations/mcp-server.md), read 10 Nov 2026.) ### Definitions, ownership and rules - Model editing tools (startDataModelEdit, writeDataModelFile, commitDataModelChanges) let an agent change definitions on a personal dev branch, merged to the default branch only with a confirm-prompting tool. (Cube's own claim; [source](https://docs.cube.dev/docs/integrations/mcp-server.md), read 10 Nov 2026.) - Query results carry the semantic layer: runQuery runs Cube SQL against views whose names and members come from searchDataModel. (Cube's own claim; [source](https://docs.cube.dev/docs/integrations/mcp-server.md), read 10 Nov 2026.) ### Freshness and upkeep - The server can target a dev-mode branch, so agents can query a branch's model before it is merged. (Cube's own claim; [source](https://docs.cube.dev/docs/integrations/mcp-server.md), read 10 Nov 2026.) - GetPreAggregationStatus and buildPreAggregation let an agent check and trigger pre-aggregation builds. (Cube's own claim; [source](https://docs.cube.dev/docs/integrations/mcp-server.md), read 10 Nov 2026.) ### Profile - Admins need admin privileges or a custom role with the Manage MCP server and connectors permission to enable MCP. (Cube's own claim; [source](https://docs.cube.dev/docs/integrations/mcp-server.md), read 10 Nov 2026.) ## Read next - [MCP servers for Cube](/mcp/cube/): 1 listed - [Cube MCP server and dbt MCP Server, side by side](/compare/cube-mcp-vs-dbt-mcp/) - [The Context Weight Index](/context-weight/) ## Sources - docs.cube.dev/docs/integrations/mcp-server.md: https://docs.cube.dev/docs/integrations/mcp-server.md (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/servers/cube-mcp/ --- # Databricks Labs MCP servers: 3 tools, remote and local, from Databricks Labs SERVER, read 10 Nov 2026 **The short answer.** Databricks Labs MCP servers is an official Labs-org MCP server for Databricks, published by Databricks Labs and read from its own source on 10 Nov 2026. It runs remote and local, and 3 tools are named. Version read: none found on 10 Nov 2026. - **Data tool:** [Databricks](/mcp/databricks/) - **Publisher:** Databricks Labs - **Kind:** Official, Labs org - **Repository or docs:** [github.com/databrickslabs/mcp](https://github.com/databrickslabs/mcp) - **Version or commit:** none found on 10 Nov 2026 - **Last release:** none found on 10 Nov 2026 - **Transport:** Remote and local - **Auth:** Databricks CLI-style credentials; Unity Catalog permissions. - **Access:** Not stated in the source read - **Tools listed:** 3 named in the evidence - **Tool definitions read at:** [raw.githubusercontent.com/databrickslabs/mcp/master/README.md](https://raw.githubusercontent.com/databrickslabs/mcp/master/README.md) - **Context weight:** Not measured: README says the tool list is inferred at startup from the Unity Catalog functions, vector search indexes and Genie spaces the user configures, so no fixed per-tool list or descriptions are published. - **Status:** Exists; README marks Unity Catalog server deprecated in favor of managed MCP servers; last push 26 Oct 2026 (GitHub API); no tagged release read. - **Read:** 10 Nov 2026 ## In one sentence The Labs server exposes Unity Catalog functions, vector search indexes and Genie spaces as tools; its README marks it deprecated in favor of the managed servers. ## What Databricks Labs MCP servers exposes 3 tools are named in the evidence read on 10 Nov 2026. `UC functions (one tool per function)`, `vector search index tools`, `Genie space tools` ## Not found in the sources read - A lineage tool in the README supported-tools list. [Source](https://raw.githubusercontent.com/databrickslabs/mcp/master/README.md). ## The evidence on Databricks Labs MCP servers, by topic ### Publisher and releases - Databrickslabs/mcp is a Databricks Labs repository whose README calls the Unity Catalog server deprecated in favor of Managed MCP servers. (Databricks Labs's own claim; [source](https://raw.githubusercontent.com/databrickslabs/mcp/master/README.md), read 10 Nov 2026.) - The repo has no GitHub release entries read; GitHub API shows last push 26 Oct 2026 and no archive flag. (Databricks Labs's own claim; [source](https://raw.githubusercontent.com/databrickslabs/mcp/master/README.md), read 10 Nov 2026.) ### Clients, auth and scoping - Tools are inferred at startup from the functions and vector search indexes in a given Unity Catalog schema plus listed Genie spaces. (Databricks Labs's own claim; [source](https://raw.githubusercontent.com/databrickslabs/mcp/master/README.md), read 10 Nov 2026.) ### Access and safety - Databricks managed MCP note in the README states permissions are always enforced for managed servers. (Databricks Labs's own claim; [source](https://raw.githubusercontent.com/databrickslabs/mcp/master/README.md), read 10 Nov 2026.) ### What it reaches - Covers Databricks Unity Catalog objects: functions, vector search indexes, Genie spaces. (Databricks Labs's own claim; [source](https://raw.githubusercontent.com/databrickslabs/mcp/master/README.md), read 10 Nov 2026.) ### Lineage and impact - The Unity Catalog server exposes UC functions, vector search indexes and Genie spaces as tools; no lineage tool. (Databricks Labs's own claim; [source](https://raw.githubusercontent.com/databrickslabs/mcp/master/README.md), read 10 Nov 2026.) ### Definitions, ownership and rules - Genie spaces and UC functions carry curated descriptions to agents; no glossary or classification tool is listed. (Databricks Labs's own claim; [source](https://raw.githubusercontent.com/databrickslabs/mcp/master/README.md), read 10 Nov 2026.) ### Freshness and upkeep - Tool list is built once at startup from the schema; no continuous refresh described. (Databricks Labs's own claim; [source](https://raw.githubusercontent.com/databrickslabs/mcp/master/README.md), read 10 Nov 2026.) ### Profile - Developer tools server in the same repo is marked under construction and not yet usable. (Databricks Labs's own claim; [source](https://raw.githubusercontent.com/databrickslabs/mcp/master/README.md), read 10 Nov 2026.) - Local use is via stdio and Databricks Apps deployment uses Streamable HTTP. (Databricks Labs's own claim; [source](https://raw.githubusercontent.com/databrickslabs/mcp/master/README.md), read 10 Nov 2026.) ## Read next - [MCP servers for Databricks](/mcp/databricks/): 2 listed - [Databricks managed MCP servers](/servers/databricks-managed-mcp/): also for Databricks - [Databricks Labs MCP servers and Databricks managed MCP servers, side by side](/compare/databrickslabs-mcp-vs-databricks-managed-mcp/) - [The Context Weight Index](/context-weight/) ## Sources - raw.githubusercontent.com/databrickslabs/mcp/master/README.md: https://raw.githubusercontent.com/databrickslabs/mcp/master/README.md (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/servers/databrickslabs-mcp/ --- # Databricks managed MCP servers: 4 tools, remote (hosted), from Databricks SERVER, read 10 Nov 2026 **The short answer.** Databricks managed MCP servers is an official MCP server for Databricks, published by Databricks and read from its own source on 10 Nov 2026. It runs remote (hosted), and 4 tools are named. Version read: Docs page last updated Nov 9, 2026. Last release: none found on 10 Nov 2026. - **Data tool:** [Databricks](/mcp/databricks/) - **Publisher:** Databricks - **Kind:** Official - **Repository or docs:** [docs.databricks.com/aws/en/generative-ai/mcp/managed-mcp](https://docs.databricks.com/aws/en/generative-ai/mcp/managed-mcp) - **Version or commit:** Docs page last updated Nov 9, 2026 - **Last release:** none found on 10 Nov 2026 - **Transport:** Remote (hosted) - **Auth:** Databricks workspace sign-in; caller's Unity Catalog permissions; Unity Gateway grants and policies. - **Access:** Configurable - **Tools listed:** 4 named in the evidence - **Tool definitions read at:** [docs.databricks.com/aws/en/generative-ai/mcp/managed-mcp](https://docs.databricks.com/aws/en/generative-ai/mcp/managed-mcp) - **Context weight:** 94 tokens for 4 tool definitions (4 with a description), measured from the docs page read - **Status:** Documented by Databricks as Databricks-provided MCPs (read 10 Nov 2026); no versioned release stream found. - **Read:** 10 Nov 2026 ## In one sentence Databricks-provided MCP endpoints run under the caller's Unity Catalog grants and include Genie spaces for questions over a governed semantic layer. ## What Databricks managed MCP servers exposes 4 tools are named in the evidence read on 10 Nov 2026. `system.ai.genie_one_mcp`, `system.ai.dbsql`, `system.ai.web_search`, `system.ai.sandbox` ## Not found in the sources read - A lineage or column-lineage MCP on the Databricks-provided MCPs page (list read). [Source](https://docs.databricks.com/aws/en/generative-ai/mcp/managed-mcp). ## The evidence on Databricks managed MCP servers, by topic ### Publisher and releases - Databricks documents Databricks-provided MCPs in the system.ai schema, with no server to host or register; the page was last updated Nov 9, 2026. (Databricks's own claim; [source](https://docs.databricks.com/aws/en/generative-ai/mcp/managed-mcp), read 10 Nov 2026.) - Genie One MCP is generally available while dbsql, sandbox and web_search require the Unity Gateway beta. (Databricks's own claim; [source](https://docs.databricks.com/aws/en/generative-ai/mcp/managed-mcp), read 10 Nov 2026.) ### Clients, auth and scoping - System.ai.genie_one_mcp answers natural-language analytics questions grounded in Genie Ontology. (Databricks's own claim; [source](https://docs.databricks.com/aws/en/generative-ai/mcp/managed-mcp), read 10 Nov 2026.) - The dbsql MCP runs SQL using the caller's Unity Catalog and warehouse permissions, so results are scoped to the asking user. (Databricks's own claim; [source](https://docs.databricks.com/aws/en/generative-ai/mcp/managed-mcp), read 10 Nov 2026.) ### Access and safety - Each system.ai MCP is a Unity Catalog securable governed by grants and per-call policies via Unity Gateway. (Databricks's own claim; [source](https://docs.databricks.com/aws/en/generative-ai/mcp/managed-mcp), read 10 Nov 2026.) - Genie Agent MCP calls bypass Unity Gateway and rely on permissions of the Genie Agent and its data. (Databricks's own claim; [source](https://docs.databricks.com/aws/en/generative-ai/mcp/managed-mcp), read 10 Nov 2026.) ### What it reaches - Beyond data, Databricks ships MCPs for Slack, GitHub, Atlassian, Google Drive, Gmail and Microsoft 365. (Databricks's own claim; [source](https://docs.databricks.com/aws/en/generative-ai/mcp/managed-mcp), read 10 Nov 2026.) ### Lineage and impact - The docs list Genie One, dbsql, web_search and sandbox MCPs; no lineage-specific MCP is listed. (Databricks's own claim; [source](https://docs.databricks.com/aws/en/generative-ai/mcp/managed-mcp), read 10 Nov 2026.) ### Definitions, ownership and rules - Genie One resolves business terms, metric definitions and table relationships through Genie Ontology, described as a governed semantic layer. (Databricks's own claim; [source](https://docs.databricks.com/aws/en/generative-ai/mcp/managed-mcp), read 10 Nov 2026.) ### Freshness and upkeep - Calls query the workspace live; no collection or refresh mechanism is described on this page. (Databricks's own claim; [source](https://docs.databricks.com/aws/en/generative-ai/mcp/managed-mcp), read 10 Nov 2026.) ### Profile - Endpoint pattern is /ai-gateway/mcp-services/. (Databricks's own claim; [source](https://docs.databricks.com/aws/en/generative-ai/mcp/managed-mcp), read 10 Nov 2026.) ## Read next - [MCP servers for Databricks](/mcp/databricks/): 2 listed - [Databricks Labs MCP servers](/servers/databrickslabs-mcp/): also for Databricks - [Databricks Labs MCP servers and Databricks managed MCP servers, side by side](/compare/databrickslabs-mcp-vs-databricks-managed-mcp/) - [The Context Weight Index](/context-weight/) ## Sources - docs.databricks.com/aws/en/generative-ai/mcp/managed-mcp: https://docs.databricks.com/aws/en/generative-ai/mcp/managed-mcp (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/servers/databricks-managed-mcp/ --- # DataHub MCP Server: 22 tools, remote and local, from Acryl Data SERVER, read 10 Nov 2026 **The short answer.** DataHub MCP Server is an official MCP server for DataHub, published by Acryl Data and read from its own source on 10 Nov 2026. It runs remote and local, and 22 tools are named. Version read: 0.8.0. Last release: 21 Oct 2026. Auth: Stdio: DATAHUB_GMS_URL + DATAHUB_GMS_TOKEN (or ~/.datahubenv). - **Data tool:** [DataHub](/mcp/datahub/) - **Publisher:** Acryl Data (DataHub) - **Kind:** Official - **Repository or docs:** [github.com/acryldata/mcp-server-datahub](https://github.com/acryldata/mcp-server-datahub) - **Version or commit:** 0.8.0 - **Last release:** 21 Oct 2026 - **Transport:** Remote and local - **Auth:** Stdio: DATAHUB_GMS_URL + DATAHUB_GMS_TOKEN (or ~/.datahubenv); HTTP: per-request DataHub personal access token as Authorization Bearer, no shared server token. - **Access:** Configurable - **Tools listed:** 22 named in the evidence - **Tool definitions read at:** [github.com/acryldata/mcp-server-datahub/blob/main/README.md](https://github.com/acryldata/mcp-server-datahub/blob/main/README.md) - **Context weight:** 567 tokens for 22 tool definitions (22 with a description), measured from the README read - **Status:** Exists under this name on 10 Nov 2026; v0.8.0 released 21 Oct 2026 per GitHub releases and PyPI. - **Read:** 10 Nov 2026 ## In one sentence DataHub's server returns column-level lineage with paths and SQL transformations, plus glossary, domain and owner discovery, with per-user tokens in HTTP mode. ## What DataHub MCP Server exposes 22 tools are named in the evidence read on 10 Nov 2026. `search`, `get_lineage`, `get_dataset_queries`, `get_entities`, `get_aspect_history`, `list_schema_fields`, `get_lineage_paths_between`, `add_tags`, `remove_tags`, `add_terms`, `remove_terms`, `add_owners`, `remove_owners`, `set_domains`, `remove_domains`, `update_description`, `add_structured_properties`, `remove_structured_properties`, `get_me`, `search_documents`, `grep_documents`, `save_document` ## Not found in the sources read - Any statement in the DataHub MCP README that tags, terms or classifications propagate along lineage when read through the MCP tools. [Source](https://raw.githubusercontent.com/acryldata/mcp-server-datahub/main/README.md). - Any freshness or ingestion-cadence statement in the DataHub MCP README (the server reads whatever DataHub has ingested). The README only describes caching of verified tokens for five minutes. [Source](https://raw.githubusercontent.com/acryldata/mcp-server-datahub/main/README.md). ## The evidence on DataHub MCP Server, by topic ### Publisher and releases - mcp-server-datahub is published from the acryldata GitHub organization (DataHub's company) and is linked from DataHub's own docs at docs.datahub.com. (observed in the source; [source](https://pypi.org/pypi/mcp-server-datahub/json), read 10 Nov 2026.) - The GitHub release page shows v0.8.0 as a published release (PyPI upload 21 Oct 2026). (observed in the source; [source](https://github.com/acryldata/mcp-server-datahub/releases/tag/v0.8.0), read 10 Nov 2026.) - The package is on PyPI as mcp-server-datahub 0.8.0. (observed in the source; [source](https://pypi.org/pypi/mcp-server-datahub/json), read 10 Nov 2026.) ### Clients, auth and scoping - The server is built to serve DataHub context to AI agents: search, lineage, query history and entity metadata. (Acryl Data's own claim; [source](https://raw.githubusercontent.com/acryldata/mcp-server-datahub/main/README.md), read 10 Nov 2026.) - In the Docker HTTP deployment each MCP client supplies its own DataHub token, so DataHub permissions and audit identity are preserved per user. (Acryl Data's own claim; [source](https://raw.githubusercontent.com/acryldata/mcp-server-datahub/main/README.md), read 10 Nov 2026.) - An optional get_me tool (TOOLS_IS_USER_ENABLED=true) returns the authenticated user's profile and group memberships. (Acryl Data's own claim; [source](https://raw.githubusercontent.com/acryldata/mcp-server-datahub/main/README.md), read 10 Nov 2026.) ### Access and safety - Mutation tools are disabled by default and need TOOLS_IS_MUTATION_ENABLED=true, so the default is read-only. (Acryl Data's own claim; [source](https://raw.githubusercontent.com/acryldata/mcp-server-datahub/main/README.md), read 10 Nov 2026.) - The HTTP entry point refuses to start if a shared server-wide token is configured, forcing per-user tokens. (Acryl Data's own claim; [source](https://raw.githubusercontent.com/acryldata/mcp-server-datahub/main/README.md), read 10 Nov 2026.) - Tokens are accepted only in the Authorization header; query-string tokens are rejected. (Acryl Data's own claim; [source](https://raw.githubusercontent.com/acryldata/mcp-server-datahub/main/README.md), read 10 Nov 2026.) ### What it reaches - The README says the server shows data in databases, lakes, warehouses and BI tools registered in DataHub; no connector list is on this page. (Acryl Data's own claim; [source](https://raw.githubusercontent.com/acryldata/mcp-server-datahub/main/README.md), read 10 Nov 2026.) ### Lineage and impact - DataHub's MCP server exposes a get_lineage tool covering datasets, columns and dashboards with hop control. (Acryl Data's own claim; [source](https://raw.githubusercontent.com/acryldata/mcp-server-datahub/main/README.md), read 10 Nov 2026.) - The README says lineage is traceable at table and column level and that transformation logic is understood. (Acryl Data's own claim; [source](https://raw.githubusercontent.com/acryldata/mcp-server-datahub/main/README.md), read 10 Nov 2026.) - A get_lineage_paths_between tool returns exact lineage paths between two assets or columns, including intermediate transformations and SQL query information. (Acryl Data's own claim; [source](https://raw.githubusercontent.com/acryldata/mcp-server-datahub/main/README.md), read 10 Nov 2026.) - Get_dataset_queries fetches real SQL queries that reference a dataset or column. (Acryl Data's own claim; [source](https://raw.githubusercontent.com/acryldata/mcp-server-datahub/main/README.md), read 10 Nov 2026.) ### Definitions, ownership and rules - DataHub's MCP server lets agents discover domains, owners, tags and glossary terms. (Acryl Data's own claim; [source](https://raw.githubusercontent.com/acryldata/mcp-server-datahub/main/README.md), read 10 Nov 2026.) - Optional mutation tools (off by default) add glossary terms, tags, owners, domains, descriptions and structured properties; this is metadata writing, and the docs do not say terms propagate along lineage through MCP. (Acryl Data's own claim; [source](https://raw.githubusercontent.com/acryldata/mcp-server-datahub/main/README.md), read 10 Nov 2026.) ### Profile - DataHub describes itself as an open-source context platform; the MCP server is Apache-2.0 licensed Python. (Acryl Data's own claim; [source](https://raw.githubusercontent.com/acryldata/mcp-server-datahub/main/README.md), read 10 Nov 2026.) ## Read next - [MCP servers for DataHub](/mcp/datahub/): 1 listed - [Atlan MCP Server and DataHub MCP Server, side by side](/compare/atlan-mcp-vs-datahub-mcp/) - [DataHub MCP Server and OpenMetadata MCP Server, side by side](/compare/datahub-mcp-vs-openmetadata-mcp/) - [DataHub MCP Server and Euno MCP, side by side](/compare/datahub-mcp-vs-euno-mcp/) - [The Context Weight Index](/context-weight/) ## Sources - raw.githubusercontent.com/acryldata/mcp-server-datahub/main/README.md: https://raw.githubusercontent.com/acryldata/mcp-server-datahub/main/README.md (read 10 Nov 2026) - pypi.org/pypi/mcp-server-datahub/json: https://pypi.org/pypi/mcp-server-datahub/json (read 10 Nov 2026) - github.com/acryldata/mcp-server-datahub/releases/tag/v0.8.0: https://github.com/acryldata/mcp-server-datahub/releases/tag/v0.8.0 (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/servers/datahub-mcp/ --- # dbt MCP Server: 52 tools, remote and local, from dbt Labs SERVER, read 10 Nov 2026 **The short answer.** dbt MCP Server is an official MCP server for dbt, published by dbt Labs and read from its own source on 10 Nov 2026. It runs remote and local, and 52 tools are named. Version read: 2.10.1. Last release: 3 Nov 2026. Auth: dbt service/personal token and environment IDs for platform tools. - **Data tool:** [dbt](/mcp/dbt/) - **Publisher:** dbt Labs - **Kind:** Official - **Repository or docs:** [github.com/dbt-labs/dbt-mcp](https://github.com/dbt-labs/dbt-mcp) - **Version or commit:** 2.10.1 - **Last release:** 3 Nov 2026 - **Transport:** Remote and local - **Auth:** dbt service/personal token and environment IDs for platform tools; OAuth or token per docs; local dbt project for CLI tools. - **Access:** Configurable - **Tools listed:** 52 named in the evidence - **Tool definitions read at:** [raw.githubusercontent.com/dbt-labs/dbt-mcp/main/README.md](https://raw.githubusercontent.com/dbt-labs/dbt-mcp/main/README.md) - **Context weight:** 1,061 tokens for 63 tool definitions (63 with a description), measured from the README read - **Status:** Exists, v2.10.1 on 3 Nov 2026 (PyPI and GitHub). - **Read:** 10 Nov 2026 ## In one sentence dbt Labs' server combines the dbt CLI, the Discovery API with model and column lineage, the Semantic Layer and SQL tools on dbt Platform. ## What dbt MCP Server exposes 52 tools are named in the evidence read on 10 Nov 2026; the tool-definition source measured for the Context Weight Index lists 63. `execute_sql`, `text_to_sql`, `get_dimension_values`, `get_dimensions`, `get_entities`, `get_metrics_compiled_sql`, `list_metrics`, `list_saved_queries`, `query_metrics`, `get_all_macros`, `get_all_models`, `get_all_sources`, `get_exposures`, `get_lineage`, `get_mart_models`, `get_model_health`, `get_model_performance`, `get_node_details`, `get_related_models`, `search_public_mesh_models`, `build`, `clone`, `compile`, `docs`, `get_lineage_dev`, `get_node_details_dev`, `list`, `parse`, `run`, `show`, `test`, `cancel_job_run`, `get_job_details`, `get_job_run_artifacts`, `get_job_run_details`, `get_job_run_error`, `list_job_run_artifacts`, `list_jobs`, `list_jobs_runs`, `list_projects`, `retry_job_run`, `trigger_job_run`, `generate_model_yaml`, `generate_source`, `generate_staging_model`, `fusion.compile_sql`, `fusion.get_column_lineage`, `get_column_lineage`, `get_product_doc_pages`, `search_product_docs`, `get_mcp_server_branch`, `get_mcp_server_version` ## Not found in the sources read No field on this entry was marked none found on 10 Nov 2026. ## The evidence on dbt MCP Server, by topic ### Publisher and releases - dbt Labs publishes dbt-mcp; the latest GitHub release is v2.10.1, whose release timestamp on the page is 3 Nov 2026 (PyPI also 3 Nov 2026). (observed in the source; [source](https://github.com/dbt-labs/dbt-mcp/releases), read 10 Nov 2026.) - dbt Labs documents the MCP server on docs.getdbt.com and ships it as self-hosted or remote. (dbt Labs's own claim; [source](https://docs.getdbt.com/docs/dbt-ai/about-mcp), read 10 Nov 2026.) ### Clients, auth and scoping - The dbt MCP server provides AI agents with context of a project in dbt Core, dbt Fusion and dbt Platform. (dbt Labs's own claim; [source](https://raw.githubusercontent.com/dbt-labs/dbt-mcp/main/README.md), read 10 Nov 2026.) ### Access and safety - dbt CLI tools can modify data models and warehouse objects, so the README warns to proceed only if you trust the client. (dbt Labs's own claim; [source](https://raw.githubusercontent.com/dbt-labs/dbt-mcp/main/README.md), read 10 Nov 2026.) - dbt docs say the server stores no production data and is a read-only access layer reading from dbt platform in real time. (dbt Labs's own claim; [source](https://docs.getdbt.com/docs/dbt-ai/about-mcp), read 10 Nov 2026.) - Configuration uses a dbt service token (DBT_TOKEN) and tool groups can be enabled or disabled by environment variables. (dbt Labs's own claim; [source](https://raw.githubusercontent.com/dbt-labs/dbt-mcp/main/.env.example), read 10 Nov 2026.) ### What it reaches - Discovery tools cover models, sources, exposures (downstream dashboards, apps) and freshness. (dbt Labs's own claim; [source](https://raw.githubusercontent.com/dbt-labs/dbt-mcp/main/README.md), read 10 Nov 2026.) ### Lineage and impact - Get_lineage returns a bounded lineage graph with type, depth and direction filters. (dbt Labs's own claim; [source](https://raw.githubusercontent.com/dbt-labs/dbt-mcp/main/README.md), read 10 Nov 2026.) - Column-level lineage tools exist: fusion.get_column_lineage via dbt Platform and get_column_lineage locally. (dbt Labs's own claim; [source](https://raw.githubusercontent.com/dbt-labs/dbt-mcp/main/README.md), read 10 Nov 2026.) - Get_lineage_dev reads lineage from the local manifest.json. (dbt Labs's own claim; [source](https://raw.githubusercontent.com/dbt-labs/dbt-mcp/main/README.md), read 10 Nov 2026.) ### Definitions, ownership and rules - Semantic Layer tools list metrics, dimensions, entities and saved queries. (dbt Labs's own claim; [source](https://raw.githubusercontent.com/dbt-labs/dbt-mcp/main/README.md), read 10 Nov 2026.) - Get_all_models returns model names and descriptions; generate_model_yaml can inherit upstream descriptions. (dbt Labs's own claim; [source](https://raw.githubusercontent.com/dbt-labs/dbt-mcp/main/README.md), read 10 Nov 2026.) ### Freshness and upkeep - Get_model_health returns run status, test results and upstream source freshness from dbt Platform. (dbt Labs's own claim; [source](https://raw.githubusercontent.com/dbt-labs/dbt-mcp/main/README.md), read 10 Nov 2026.) ### Profile - The README publishes an experimental MCP bundle (dbt-mcp.mcpb) with each release. (dbt Labs's own claim; [source](https://raw.githubusercontent.com/dbt-labs/dbt-mcp/main/README.md), read 10 Nov 2026.) ## Read next - [MCP servers for dbt](/mcp/dbt/): 1 listed - [Cube MCP server and dbt MCP Server, side by side](/compare/cube-mcp-vs-dbt-mcp/) - [The Context Weight Index](/context-weight/) ## Sources - github.com/dbt-labs/dbt-mcp/releases: https://github.com/dbt-labs/dbt-mcp/releases (read 10 Nov 2026) - docs.getdbt.com/docs/dbt-ai/about-mcp: https://docs.getdbt.com/docs/dbt-ai/about-mcp (read 10 Nov 2026) - raw.githubusercontent.com/dbt-labs/dbt-mcp/main/README.md: https://raw.githubusercontent.com/dbt-labs/dbt-mcp/main/README.md (read 10 Nov 2026) - raw.githubusercontent.com/dbt-labs/dbt-mcp/main/.env.example: https://raw.githubusercontent.com/dbt-labs/dbt-mcp/main/.env.example (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/servers/dbt-mcp/ --- # Euno MCP: 56 tools, remote (hosted), from Euno SERVER, read 10 Nov 2026 **The short answer.** Euno MCP is an official MCP server for Euno, published by Euno and read from its own source on 10 Nov 2026. It runs remote (hosted), and 56 tools are named. Version read: Hosted service at api.app.euno.ai/mcp, no version published (docs read 10 Nov 2026). - **Data tool:** [Euno](/mcp/euno/) - **Publisher:** Euno - **Kind:** Official - **Repository or docs:** [docs.euno.ai/ai/mcp-setup](https://docs.euno.ai/ai/mcp-setup?utm_source=datamcpservers.com&utm_medium=referral&utm_campaign=server) - **Version or commit:** Hosted service at api.app.euno.ai/mcp, no version published (docs read 10 Nov 2026) - **Last release:** none found on 10 Nov 2026 - **Transport:** Remote (hosted) - **Auth:** OAuth sign-in (default) or API key and headers; active persona scopes data and actions. - **Access:** Configurable - **Tools listed:** 56 named in the evidence - **Tool definitions read at:** [docs.euno.ai/ai/mcp-setup.md](https://docs.euno.ai/ai/mcp-setup.md?utm_source=datamcpservers.com&utm_medium=referral&utm_campaign=server) - **Context weight:** 760 tokens for 56 tool definitions (42 with a description), measured from the docs page read - **Status:** Exists on 10 Nov 2026; documented in docs.euno.ai MCP Setup Guide; tool names and descriptions published, parameter schemas not published. - **Read:** 10 Nov 2026 ## In one sentence Euno's hosted OAuth endpoint serves persona-scoped lineage, impact analysis, EQL search and catalog tools over its metadata graph, with no warehouse query tool. ## What Euno MCP exposes 56 tools are named in the evidence read on 10 Nov 2026. `load_account_settings`, `list_personas`, `current_persona`, `switch_persona`, `load_skill`, `search_properties`, `fetch_single_resource`, `find_resource_by_name`, `find_resources_for_topic`, `get_upstream_lineage`, `resource_impact_analysis`, `query_to_eql`, `search_resources_eql`, `count_resources_eql`, `multi_count_resources_eql`, `pivot_resources_eql`, `search_resource_audit_log`, `sensitive_data_incident_response`, `documentation_search`, `documentation_get_full_document`, `documentation_get_surrounding_context`, `list_metadata_tags`, `list_metadata_tag_domains`, `preview_metadata_tag_plan`, `get_context_benchmark_overview`, `get_context_benchmark_questions`, `list_context_benchmark_runs`, `get_context_benchmark_run`, `get_context_benchmark_improvement_report`, `get_context_benchmark_question_result`, `get_context_benchmark_suggestion_application`, `get_context_benchmark_schedule`, `create_context_benchmark_question`, `update_context_benchmark_question`, `delete_context_benchmark_question`, `start_context_benchmark_run`, `stop_context_benchmark_run`, `update_context_benchmark_schedule`, `prepare_context_benchmark_suggestion_application`, `decide_context_benchmark_suggestion_application`, `undo_context_benchmark_suggestion`, `reapply_context_benchmark_suggestion`, `fetch_glossary_term`, `find_glossary_term`, `create_glossary_term`, `update_glossary_term`, `delete_glossary_term`, `fetch_metadata_tag`, `create_metadata_tag`, `update_metadata_tag`, `delete_metadata_tag`, `fetch_workflow`, `find_workflow`, `create_workflow`, `update_workflow`, `delete_workflow` ## Not found in the sources read - A version number or dated release list for the Euno MCP server; the docs describe a hosted service, and the separate Euno SDK is v0.5. [Source](https://docs.euno.ai/ai/mcp-setup.md?utm_source=datamcpservers.com&utm_medium=referral&utm_campaign=server). **What Euno's server does not do.** Euno's server gives an agent governed context about the data; it does not run the query, which is the job of the warehouse's own server. ## The evidence on Euno MCP, by topic ### Publisher and releases - Euno publishes and documents its own MCP server at api.app.euno.ai/mcp. (Euno's own claim; [source](https://docs.euno.ai/ai/mcp-setup.md?utm_source=datamcpservers.com&utm_medium=referral&utm_campaign=server), read 10 Nov 2026.) - Stdio use goes through the third-party open-source mcp-remote bridge, not a Euno-published package. (Euno's own claim; [source](https://docs.euno.ai/ai/mcp-setup/mcp-bridge-setup.md?utm_source=datamcpservers.com&utm_medium=referral&utm_campaign=server), read 10 Nov 2026.) ### Clients, auth and scoping - Euno MCP uses the active persona; if none can be identified the user is asked to choose. (Euno's own claim; [source](https://docs.euno.ai/ai/mcp-setup.md?utm_source=datamcpservers.com&utm_medium=referral&utm_campaign=server), read 10 Nov 2026.) - Euno MCP works with Cursor, Claude, ChatGPT in Developer Mode and Claude Code. (Euno's own claim; [source](https://docs.euno.ai/ai/mcp-setup.md?utm_source=datamcpservers.com&utm_medium=referral&utm_campaign=server), read 10 Nov 2026.) - Euno MCP has a dedicated Gemini Enterprise setup path besides direct, bridge and API-key setups. (Euno's own claim; [source](https://docs.euno.ai/ai/mcp-setup.md?utm_source=datamcpservers.com&utm_medium=referral&utm_campaign=server), read 10 Nov 2026.) ### Access and safety - Euno MCP includes read and write tools; write tools such as catalog creates and benchmark changes ask for client confirmation. (Euno's own claim; [source](https://docs.euno.ai/ai/mcp-setup.md?utm_source=datamcpservers.com&utm_medium=referral&utm_campaign=server), read 10 Nov 2026.) - Euno MCP catalog reads are published read-only and catalog writes are marked destructive; deletes need confirm=true. (Euno's own claim; [source](https://docs.euno.ai/ai/mcp-setup.md?utm_source=datamcpservers.com&utm_medium=referral&utm_campaign=server), read 10 Nov 2026.) - Euno enforces authentication and persona permissions on the server, independent of client confirmation. (Euno's own claim; [source](https://docs.euno.ai/ai/mcp-setup.md?utm_source=datamcpservers.com&utm_medium=referral&utm_campaign=server), read 10 Nov 2026.) - Context benchmark tools are accepted only for an account admin persona. (Euno's own claim; [source](https://docs.euno.ai/ai/mcp-setup.md?utm_source=datamcpservers.com&utm_medium=referral&utm_campaign=server), read 10 Nov 2026.) ### What it reaches - The published tool table covers metadata, lineage, EQL, audit, documentation, catalog and benchmark tools; no listed tool executes SQL against a warehouse, and query_to_eql converts a filter without executing it. (observed in the source; [source](https://docs.euno.ai/ai/mcp-setup.md?utm_source=datamcpservers.com&utm_medium=referral&utm_campaign=server), read 10 Nov 2026.) - Euno's FAQ says Euno does not query customer data. (Euno's own claim; [source](https://euno.ai/?utm_source=datamcpservers.com&utm_medium=referral&utm_campaign=server), read 10 Nov 2026.) ### Lineage and impact - Euno MCP exposes get_upstream_lineage and resource_impact_analysis tools for one resource URI. (Euno's own claim; [source](https://docs.euno.ai/ai/mcp-setup.md?utm_source=datamcpservers.com&utm_medium=referral&utm_campaign=server), read 10 Nov 2026.) - Euno MCP can search and aggregate the metadata graph with EQL (search, count, multi-count, pivot). (Euno's own claim; [source](https://docs.euno.ai/ai/mcp-setup.md?utm_source=datamcpservers.com&utm_medium=referral&utm_campaign=server), read 10 Nov 2026.) - Euno MCP's example prompt traces a Tableau dashboard's upstream lineage. (Euno's own claim; [source](https://docs.euno.ai/ai/mcp-setup.md?utm_source=datamcpservers.com&utm_medium=referral&utm_campaign=server), read 10 Nov 2026.) ### Definitions, ownership and rules - Euno MCP has glossary, metadata tag and workflow tools (14 direct catalog tools). (Euno's own claim; [source](https://docs.euno.ai/ai/mcp-setup.md?utm_source=datamcpservers.com&utm_medium=referral&utm_campaign=server), read 10 Nov 2026.) - Euno MCP's load_account_settings returns account, persona, skill and metadata context; skills carry organization-specific workflows. (Euno's own claim; [source](https://docs.euno.ai/ai/mcp-setup.md?utm_source=datamcpservers.com&utm_medium=referral&utm_campaign=server), read 10 Nov 2026.) - Euno MCP includes sensitive_data_incident_response returning evidence per user and BI-resource pair. (Euno's own claim; [source](https://docs.euno.ai/ai/mcp-setup.md?utm_source=datamcpservers.com&utm_medium=referral&utm_campaign=server), read 10 Nov 2026.) ### Freshness and upkeep - The Euno MCP server keeps no session state, so connections survive restarts and scaling. (Euno's own claim; [source](https://docs.euno.ai/ai/mcp-setup.md?utm_source=datamcpservers.com&utm_medium=referral&utm_campaign=server), read 10 Nov 2026.) - Euno's October 2026 release notes add pivot and multi-count aggregation tools through MCP. (Euno's own claim; [source](https://docs.euno.ai/support-resources/release-note/october-2026.md?utm_source=datamcpservers.com&utm_medium=referral&utm_campaign=server), read 10 Nov 2026.) ### Profile - Euno's hosted MCP uses OAuth; an API-key setup exists for clients that cannot do interactive sign-in. (Euno's own claim; [source](https://docs.euno.ai/ai/mcp-api-key-setup.md?utm_source=datamcpservers.com&utm_medium=referral&utm_campaign=server), read 10 Nov 2026.) ## Read next - [MCP servers for Euno](/mcp/euno/): 1 listed - [Atlan MCP Server and Euno MCP, side by side](/compare/atlan-mcp-vs-euno-mcp/) - [DataHub MCP Server and Euno MCP, side by side](/compare/datahub-mcp-vs-euno-mcp/) - [Collibra MCP Server (chip) and Euno MCP, side by side](/compare/collibra-mcp-vs-euno-mcp/) - [The Context Weight Index](/context-weight/) ## Sources - docs.euno.ai/ai/mcp-setup.md: https://docs.euno.ai/ai/mcp-setup.md (read 10 Nov 2026) - euno.ai: https://euno.ai/ (read 10 Nov 2026) - docs.euno.ai/support-resources/release-note/october-2026.md: https://docs.euno.ai/support-resources/release-note/october-2026.md (read 10 Nov 2026) - docs.euno.ai/ai/mcp-setup/mcp-bridge-setup.md: https://docs.euno.ai/ai/mcp-setup/mcp-bridge-setup.md (read 10 Nov 2026) - docs.euno.ai/ai/mcp-api-key-setup.md: https://docs.euno.ai/ai/mcp-api-key-setup.md (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/servers/euno-mcp/ --- # Fivetran MCP Server: 53 tools, remote and local, from Fivetran SERVER, read 10 Nov 2026 **The short answer.** Fivetran MCP Server is an official MCP server for Fivetran, published by Fivetran and read from its own source on 10 Nov 2026. It runs remote and local, and 53 tools are named. Version read: V0.5.1. Last release: 6 Oct 2026. Auth: Fivetran API key and API secret (FIVETRAN_API_KEY, FIVETRAN_API_SECRET) from the Fivetran dashboard. - **Data tool:** [Fivetran](/mcp/fivetran/) - **Publisher:** Fivetran - **Kind:** Official - **Repository or docs:** [github.com/fivetran/fivetran-mcp](https://github.com/fivetran/fivetran-mcp) - **Version or commit:** V0.5.1 - **Last release:** 6 Oct 2026 - **Transport:** Remote and local - **Auth:** Fivetran API key and API secret (FIVETRAN_API_KEY, FIVETRAN_API_SECRET) from the Fivetran dashboard; one shared credential, not per-user OAuth. Streamable HTTP mode forwards headers; OAuth resource-server support is scaffolding only. - **Access:** Configurable - **Tools listed:** 53 named in the evidence - **Tool definitions read at:** [raw.githubusercontent.com/fivetran/fivetran-mcp/main/open-api-definitions/AVAILABLE_ACTIONS.md](https://raw.githubusercontent.com/fivetran/fivetran-mcp/main/open-api-definitions/AVAILABLE_ACTIONS.md) - **Context weight:** 299 tokens for 53 tool definitions (2 with a description), measured from the README read - **Status:** Exists on 10 Nov 2026 as fivetran/fivetran-mcp, installed from git with uvx (the PyPI names fivetran-mcp and mcp-fivetran belong to unrelated projects). Tool set is generated from the OpenAPI manifest, so it mirrors the REST API. - **Read:** 10 Nov 2026 ## In one sentence Fivetran's server mirrors its REST API behind a read-by-default scope with a deny-list, so an agent can inspect connections, schemas and sync status. ## What Fivetran MCP Server exposes 53 tools are named in the evidence read on 10 Nov 2026. `list_endpoints`, `get_schema`, `account_read`, `connections_delete`, `connections_read`, `connections_write`, `connector_sdk_delete`, `connector_sdk_read`, `connector_sdk_write`, `destinations_delete`, `destinations_read`, `destinations_write`, `external_logging_delete`, `external_logging_read`, `external_logging_write`, `external_secrets_managers_delete`, `external_secrets_managers_read`, `external_secrets_managers_write`, `external_secrets_managers_entities_read`, `groups_delete`, `groups_read`, `groups_write`, `hvr_write`, `hybrid_deployment_agents_delete`, `hybrid_deployment_agents_read`, `hybrid_deployment_agents_write`, `metadata_read`, `private_links_delete`, `private_links_read`, `private_links_write`, `proxy_delete`, `proxy_read`, `proxy_write`, `public_read`, `roles_read`, `system_keys_delete`, `system_keys_read`, `system_keys_write`, `teams_delete`, `teams_read`, `teams_write`, `transformation_projects_delete`, `transformation_projects_read`, `transformation_projects_write`, `transformations_delete`, `transformations_read`, `transformations_write`, `users_delete`, `users_read`, `users_write`, `webhooks_delete`, `webhooks_read`, `webhooks_write` ## Not found in the sources read - Any lineage path in the OpenAPI definition the server loads (searched fivetran-open-api-definition.json for 'lineage' in endpoint paths). [Source](https://raw.githubusercontent.com/fivetran/fivetran-mcp/main/fivetran-open-api-definition.json). - Glossary, ownership or classification tools in the Fivetran MCP README or AVAILABLE_ACTIONS.md. [Source](https://raw.githubusercontent.com/fivetran/fivetran-mcp/main/README.md). ## The evidence on Fivetran MCP Server, by topic ### Publisher and releases - Fivetran publishes the server in its own GitHub organization and describes it as an MCP server for interacting with a Fivetran environment. (Fivetran's own claim; [source](https://raw.githubusercontent.com/fivetran/fivetran-mcp/main/README.md), read 10 Nov 2026.) - The latest GitHub release is v0.5.1, timestamped 6 Oct 2026 on the releases page. (observed in the source; [source](https://github.com/fivetran/fivetran-mcp/releases), read 10 Nov 2026.) - The README says the PyPI names fivetran-mcp and mcp-fivetran belong to unrelated projects, so the server is run from the git URL and has no versioned PyPI package. (Fivetran's own claim; [source](https://raw.githubusercontent.com/fivetran/fivetran-mcp/main/README.md), read 10 Nov 2026.) ### Clients, auth and scoping - The server uses one Fivetran API key and secret, so agent actions run with that credential's scope instead of the asking user's identity. (Fivetran's own claim; [source](https://raw.githubusercontent.com/fivetran/fivetran-mcp/main/README.md), read 10 Nov 2026.) - The README states OAuth resource-server support in streamable-http mode is not yet functional. (Fivetran's own claim; [source](https://raw.githubusercontent.com/fivetran/fivetran-mcp/main/README.md), read 10 Nov 2026.) - Two discovery tools plus one execution tool per allowed resource/action pair keep the tool list small and filtered by scope. (Fivetran's own claim; [source](https://raw.githubusercontent.com/fivetran/fivetran-mcp/main/README.md), read 10 Nov 2026.) ### Access and safety - FIVETRAN_SCOPE defaults to read and can be raised to read/write or read/write/delete. (Fivetran's own claim; [source](https://raw.githubusercontent.com/fivetran/fivetran-mcp/main/README.md), read 10 Nov 2026.) - DISALLOWED_ACTIONS denies resource:action pairs or single endpoints, and the README advises a denylist for credential endpoints because they outlive the session. (Fivetran's own claim; [source](https://raw.githubusercontent.com/fivetran/fivetran-mcp/main/README.md), read 10 Nov 2026.) - Write and delete confirmation is advisory only; the server does not enforce it. (Fivetran's own claim; [source](https://raw.githubusercontent.com/fivetran/fivetran-mcp/main/README.md), read 10 Nov 2026.) ### What it reaches - Resource groups include connections, destinations, groups, external-logging, external-secrets-managers, hybrid-deployment-agents, connector-sdk and account. (observed in the source; [source](https://raw.githubusercontent.com/fivetran/fivetran-mcp/main/open-api-definitions/AVAILABLE_ACTIONS.md), read 10 Nov 2026.) - The server runs over stdio by default and can run as a hosted HTTP server with --transport streamable-http. (Fivetran's own claim; [source](https://raw.githubusercontent.com/fivetran/fivetran-mcp/main/README.md), read 10 Nov 2026.) ### Lineage and impact - The server exposes every endpoint in the Fivetran API through generated tools; connection schema, table and column configuration endpoints are in the OpenAPI manifest (configuration level, not lineage). (observed in the source; [source](https://raw.githubusercontent.com/fivetran/fivetran-mcp/main/fivetran-open-api-definition.json), read 10 Nov 2026.) - Get_schema returns the parameters, request body and response schema of an endpoint, and for connection create/modify it can add service-specific configuration fields. (Fivetran's own claim; [source](https://raw.githubusercontent.com/fivetran/fivetran-mcp/main/README.md), read 10 Nov 2026.) ### Definitions, ownership and rules - The only metadata resource in the action list is metadata:read, which maps to the connector-types endpoints (/v1/metadata/connector-types). (observed in the source; [source](https://raw.githubusercontent.com/fivetran/fivetran-mcp/main/fivetran-open-api-definition.json), read 10 Nov 2026.) ### Freshness and upkeep - The README's example questions are about live pipeline state, such as the last time a postgres connection completed a sync and whether any connections are broken. (Fivetran's own claim; [source](https://raw.githubusercontent.com/fivetran/fivetran-mcp/main/README.md), read 10 Nov 2026.) - The OpenAPI manifest includes sync-history, state and sync endpoints per connection. (observed in the source; [source](https://raw.githubusercontent.com/fivetran/fivetran-mcp/main/fivetran-open-api-definition.json), read 10 Nov 2026.) ### Profile - The README lists a copy-connections plugin for Claude Code and Codex that uses this server, hosted in its own repository. (Fivetran's own claim; [source](https://raw.githubusercontent.com/fivetran/fivetran-mcp/main/README.md), read 10 Nov 2026.) - Setup requires Python 3.10+ and uv; API credentials are generated in the Fivetran dashboard. (Fivetran's own claim; [source](https://raw.githubusercontent.com/fivetran/fivetran-mcp/main/README.md), read 10 Nov 2026.) ## Read next - [MCP servers for Fivetran](/mcp/fivetran/): 1 listed - [Airbyte MCP (hosted) and Fivetran MCP Server, side by side](/compare/airbyte-mcp-vs-fivetran-mcp/) - [The Context Weight Index](/context-weight/) ## Sources - raw.githubusercontent.com/fivetran/fivetran-mcp/main/README.md: https://raw.githubusercontent.com/fivetran/fivetran-mcp/main/README.md (read 10 Nov 2026) - github.com/fivetran/fivetran-mcp/releases: https://github.com/fivetran/fivetran-mcp/releases (read 10 Nov 2026) - raw.githubusercontent.com/fivetran/fivetran-mcp/main/fivetran-open-api-definition.json: https://raw.githubusercontent.com/fivetran/fivetran-mcp/main/fivetran-open-api-definition.json (read 10 Nov 2026) - raw.githubusercontent.com/fivetran/fivetran-mcp/main/open-api-definitions/AVAILABLE_ACTIONS.md: https://raw.githubusercontent.com/fivetran/fivetran-mcp/main/open-api-definitions/AVAILABLE_ACTIONS.md (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/servers/fivetran-mcp/ --- # Hex MCP server: 18 tools, remote (hosted), from Hex SERVER, read 10 Nov 2026 **The short answer.** Hex MCP server is an official MCP server for Hex, published by Hex and read from its own source on 10 Nov 2026. It runs remote (hosted), and 18 tools are named. Version read: Hosted service (no versioned release). Last release: none found on 10 Nov 2026. - **Data tool:** [Hex](/mcp/hex/) - **Publisher:** Hex - **Kind:** Official - **Repository or docs:** [learn.hex.tech/docs/api-integrations/mcp-server](https://learn.hex.tech/docs/api-integrations/mcp-server) - **Version or commit:** Hosted service (no versioned release) - **Last release:** none found on 10 Nov 2026 - **Transport:** Remote (hosted) - **Auth:** OAuth authorization to the user's Hex workspace (server URL app.hex.tech/mcp, or the customer's single-tenant, EU or HIPAA domain); requires Explorer role or higher. - **Access:** Read and write - **Tools listed:** 18 named in the evidence - **Tool definitions read at:** [learn.hex.tech/docs/api-integrations/mcp-server](https://learn.hex.tech/docs/api-integrations/mcp-server) - **Context weight:** 252 tokens for 18 tool definitions (18 with a description), measured from the docs page read - **Status:** Exists on 10 Nov 2026 as a Hex-hosted MCP server; tool list read from the Hex docs page (knowledge tools for Explorers, project editing tools for Editors). - **Read:** 10 Nov 2026 ## In one sentence Hex's hosted server runs Threads (agent analyses over the user's data connections) and notebook editing under workspace roles and project permissions. ## What Hex MCP server exposes 18 tools are named in the evidence read on 10 Nov 2026. `search_projects`, `create_thread`, `get_thread`, `continue_thread`, `get_me`, `create_project`, `get_project`, `list_cells`, `get_cell`, `create_cell`, `update_cell`, `delete_cell`, `run_cell`, `run_notebook`, `get_run`, `get_cell_output`, `get_cell_image`, `list_data_connections` ## Not found in the sources read - A lineage tool on the Hex MCP server page (cell-level reads exist, but no column or upstream-table lineage tool is listed). [Source](https://learn.hex.tech/docs/api-integrations/mcp-server). - Scheduled metadata collection on the Hex MCP page. [Source](https://learn.hex.tech/docs/api-integrations/mcp-server). ## The evidence on Hex MCP server, by topic ### Publisher and releases - Hex documents its own MCP server and an official Hex plugin in the Cursor Marketplace. (Hex's own claim; [source](https://learn.hex.tech/docs/api-integrations/mcp-server), read 10 Nov 2026.) - No versioned release is published; the hosted URL is app.hex.tech/mcp. (Hex's own claim; [source](https://learn.hex.tech/docs/api-integrations/mcp-server), read 10 Nov 2026.) ### Clients, auth and scoping - Threads created through the MCP server behave like Threads in the Hex app and select among the data connections the user can access. (Hex's own claim; [source](https://learn.hex.tech/docs/api-integrations/mcp-server), read 10 Nov 2026.) - The MCP server enforces workspace roles and project permissions; editing a cell needs Can Edit or Full Access on that project, even for Admins. (Hex's own claim; [source](https://learn.hex.tech/docs/api-integrations/mcp-server), read 10 Nov 2026.) - Knowledge tools need the Explorer role or above and project editing tools need Editor or above. (Hex's own claim; [source](https://learn.hex.tech/docs/api-integrations/mcp-server), read 10 Nov 2026.) ### Access and safety - The MCP server does not query Sensitive data connections by default and cannot read cell outputs from projects that use them, unless an admin allows it. (Hex's own claim; [source](https://learn.hex.tech/docs/api-integrations/mcp-server), read 10 Nov 2026.) - MCP access is not governed by the workspace API access setting; deactivating the user is how to revoke all of a user's access. (Hex's own claim; [source](https://learn.hex.tech/docs/api-integrations/mcp-server), read 10 Nov 2026.) - The tools include create, update and delete for cells, so the server is read-write for Editors. (Hex's own claim; [source](https://learn.hex.tech/docs/api-integrations/mcp-server), read 10 Nov 2026.) ### What it reaches - The server works with Claude Desktop, Claude Code, Cursor, ChatGPT, Codex and Glean as well as standard MCP clients. (Hex's own claim; [source](https://learn.hex.tech/docs/api-integrations/mcp-server), read 10 Nov 2026.) - Single-tenant, EU multi-tenant and HIPAA multi-tenant workspaces use their own host instead of app.hex.tech. (Hex's own claim; [source](https://learn.hex.tech/docs/api-integrations/mcp-server), read 10 Nov 2026.) ### Lineage and impact - Project editing tools list cells and read cell outputs and chart images, giving agents the SQL, Python and Markdown cells of a notebook. (Hex's own claim; [source](https://learn.hex.tech/docs/api-integrations/mcp-server), read 10 Nov 2026.) - Search_projects finds projects in the workspace, and get_project reads a project's metadata. (Hex's own claim; [source](https://learn.hex.tech/docs/api-integrations/mcp-server), read 10 Nov 2026.) ### Definitions, ownership and rules - Admins can mark data connections Sensitive and the docs point to descriptions and exclusions as the way to shape what the Hex Agent sees. (Hex's own claim; [source](https://learn.hex.tech/docs/api-integrations/mcp-server), read 10 Nov 2026.) ### Freshness and upkeep - Run_cell and run_notebook execute cells or the draft notebook and get_run polls status, so results are produced at call time. (Hex's own claim; [source](https://learn.hex.tech/docs/api-integrations/mcp-server), read 10 Nov 2026.) ### Profile - The llms.txt notes workspace data is private and MCP access requires the user to authorize an OAuth connection. (Hex's own claim; [source](https://learn.hex.tech/llms.txt), read 10 Nov 2026.) - Claude connector widgets created by the Hex connector expire after 24 hours, and file uploads to a conversation are not supported. (Hex's own claim; [source](https://learn.hex.tech/docs/api-integrations/mcp-server), read 10 Nov 2026.) ## Read next - [MCP servers for Hex](/mcp/hex/): 1 listed - [Hex MCP server and Omni MCP Server, side by side](/compare/hex-mcp-vs-omni-mcp/) - [The Context Weight Index](/context-weight/) ## Sources - learn.hex.tech/docs/api-integrations/mcp-server: https://learn.hex.tech/docs/api-integrations/mcp-server (read 10 Nov 2026) - learn.hex.tech/llms.txt: https://learn.hex.tech/llms.txt (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/servers/hex-mcp/ --- # Looker tools in MCP Toolbox for Databases: 55 tools, remote and local, from Google SERVER, read 10 Nov 2026 **The short answer.** Looker tools in MCP Toolbox for Databases is an official MCP server for Looker, published by Google and read from its own source on 10 Nov 2026. It runs remote and local, and 55 tools are named. Version read: V1.16.0. Last release: 5 Nov 2026. - **Data tool:** [Looker](/mcp/looker/) - **Publisher:** Google (googleapis/mcp-toolbox) - **Kind:** Official - **Repository or docs:** [github.com/googleapis/mcp-toolbox](https://github.com/googleapis/mcp-toolbox) - **Version or commit:** V1.16.0 - **Last release:** 5 Nov 2026 - **Transport:** Remote and local - **Auth:** Looker API client_id and client_secret in the Looker source config, or use_client_oauth to pass the caller's OAuth token (default false). - **Access:** Configurable - **Tools listed:** 55 named in the evidence - **Tool definitions read at:** [mcp-toolbox.dev/llms.txt](https://mcp-toolbox.dev/llms.txt) - **Context weight:** 1,475 tokens for 55 tool definitions (55 with a description), measured from the docs index read - **Status:** Exists on 10 Nov 2026 as the looker source type in MCP Toolbox for Databases; the repo URL github.com/googleapis/genai-toolbox now titles itself googleapis/mcp-toolbox. 55 looker-* tools are listed in the Toolbox llms.txt index; schema file holds the one-line descriptions only. - **Read:** 10 Nov 2026 ## In one sentence Google's Toolbox exposes Looker models, explores, dimensions, measures and LookML project files, plus instance health and usage tools. ## What Looker tools in MCP Toolbox for Databases exposes 55 tools are named in the evidence read on 10 Nov 2026. `looker-add-dashboard-element`, `looker-add-dashboard-filter`, `looker-conversational-analytics`, `looker-create-agent`, `looker-create-dashboard-layout`, `looker-create-git-branch`, `looker-create-project-directory`, `looker-create-project-file`, `looker-delete-agent`, `looker-delete-git-branch`, `looker-delete-project-directory`, `looker-delete-project-file`, `looker-dev-mode`, `looker-generate-embed-url`, `looker-get-agent`, `looker-get-connection-databases`, `looker-get-connection-schemas`, `looker-get-connection-table-columns`, `looker-get-connection-tables`, `looker-get-connections`, `looker-get-dashboard`, `looker-get-dashboards`, `looker-get-dimensions`, `looker-get-explore`, `looker-get-explores`, `looker-get-field-value-suggestions`, `looker-get-filters`, `looker-get-git-branch`, `looker-get-looks`, `looker-get-measures`, `looker-get-models`, `looker-get-parameters`, `looker-get-project-directories`, `looker-get-project-file`, `looker-get-project-files`, `looker-get-projects`, `looker-health-analyze`, `looker-health-pulse`, `looker-health-vacuum`, `looker-import-dashboard-from-lookml`, `looker-list-agents`, `looker-list-git-branches`, `looker-make-dashboard`, `looker-make-look`, `looker-query`, `looker-query-sql`, `looker-query-url`, `looker-run-dashboard`, `looker-run-look`, `looker-switch-git-branch`, `looker-update-agent`, `looker-update-dashboard-element`, `looker-update-dashboard-layout-component`, `looker-update-project-file`, `looker-validate-project` ## Not found in the sources read - A lineage tool among the 55 looker-* tools in the Toolbox llms.txt index (no column-lineage or upstream-table tool is named). [Source](https://mcp-toolbox.dev/llms.txt). - Scheduled or event-driven metadata collection for Looker in the Toolbox docs; tools call the Looker API per request. [Source](https://mcp-toolbox.dev/integrations/looker/source/). ## The evidence on Looker tools in MCP Toolbox for Databases, by topic ### Publisher and releases - Looker's tools ship in MCP Toolbox for Databases, published under Google's googleapis GitHub organization. (Google's own claim; [source](https://raw.githubusercontent.com/googleapis/genai-toolbox/main/README.md), read 10 Nov 2026.) - The latest GitHub release is v1.16.0, timestamped 5 Nov 2026; the releases page title shows the repo as googleapis/mcp-toolbox. (observed in the source; [source](https://github.com/googleapis/mcp-toolbox/releases), read 10 Nov 2026.) - The release list shows v1.16.0, v1.15.0, v1.14.0 and v1.13.1 as the four most recent tags. (observed in the source; [source](https://github.com/googleapis/mcp-toolbox/releases), read 10 Nov 2026.) ### Clients, auth and scoping - The Looker source can pass the caller's OAuth token instead of one service credential via use_client_oauth, which defaults to false. (Google's own claim; [source](https://mcp-toolbox.dev/integrations/looker/source/), read 10 Nov 2026.) - The default source config authenticates with a single client_id and client_secret, so identity is that API user unless OAuth is enabled. (Google's own claim; [source](https://mcp-toolbox.dev/integrations/looker/source/), read 10 Nov 2026.) - The source can hide models, explores and fields from agents with show_hidden_models, show_hidden_explores and show_hidden_fields (all default true). (Google's own claim; [source](https://mcp-toolbox.dev/integrations/looker/source/), read 10 Nov 2026.) ### Access and safety - Toolbox documents a Read-Only Tools page covering engine-level protocol locks, tool suppression and MCP annotations. (Google's own claim; [source](https://mcp-toolbox.dev/llms.txt), read 10 Nov 2026.) - The Looker toolset includes write tools such as creating or deleting LookML project files and switching Looker into dev mode, so access is not read-only by default. (Google's own claim; [source](https://mcp-toolbox.dev/llms.txt), read 10 Nov 2026.) - Toolbox documents screening traffic with Google Cloud Model Armor to protect against prompt injection and sensitive data leakage. (Google's own claim; [source](https://mcp-toolbox.dev/llms.txt), read 10 Nov 2026.) - Toolbox includes OpenTelemetry observability built in. (Google's own claim; [source](https://raw.githubusercontent.com/googleapis/genai-toolbox/main/README.md), read 10 Nov 2026.) ### What it reaches - Looker can be deployed in the cloud, on GCP, or on premises, and the same Toolbox covers Google Cloud databases such as AlloyDB, BigQuery, Cloud SQL, Spanner and Firestore. (Google's own claim; [source](https://mcp-toolbox.dev/integrations/looker/source/), read 10 Nov 2026.) - Looker connection tools list databases, schemas, tables and columns of the warehouse connections configured in Looker. (Google's own claim; [source](https://mcp-toolbox.dev/llms.txt), read 10 Nov 2026.) - Toolbox runs as a local stdio MCP server through npx @toolbox-sdk/server with the --stdio flag. (Google's own claim; [source](https://github.com/googleapis/mcp-toolbox), read 10 Nov 2026.) ### Lineage and impact - Looker-get-explore returns explore metadata including always_filter and conditionally_filter constraints, tags and hidden status. (Google's own claim; [source](https://mcp-toolbox.dev/integrations/looker/tools/looker-get-explore/), read 10 Nov 2026.) - Separate tools return dimensions, measures, filters and parameters of an explore, and the columns of connection tables. (Google's own claim; [source](https://mcp-toolbox.dev/llms.txt), read 10 Nov 2026.) - Looker-get-project-file returns the contents of a LookML file, so an agent can read the transformation and field logic written in LookML. (Google's own claim; [source](https://mcp-toolbox.dev/llms.txt), read 10 Nov 2026.) ### Definitions, ownership and rules - The Looker source page describes Looker as providing a semantic layer to facilitate querying. (Google's own claim; [source](https://mcp-toolbox.dev/integrations/looker/source/), read 10 Nov 2026.) - Looker-health-analyze reports unused explores, joins and fields, defining unused as not used as a field or filter in the past 90 days in production. (Google's own claim; [source](https://mcp-toolbox.dev/integrations/looker/tools/looker-health-analyze/), read 10 Nov 2026.) - Looker-conversational-analytics uses the Conversational Analytics API to analyze Looker data, and agents can be created and managed with looker-create-agent. (Google's own claim; [source](https://mcp-toolbox.dev/llms.txt), read 10 Nov 2026.) ### Freshness and upkeep - Looker-health-pulse performs health checks on a Looker instance, including database connections and dashboard performance, at call time. (Google's own claim; [source](https://mcp-toolbox.dev/llms.txt), read 10 Nov 2026.) ### Profile - The repo was renamed: the genai-toolbox URL now serves googleapis/mcp-toolbox, and prebuilt tools exist for Looker alongside many Google databases. (Google's own claim; [source](https://mcp-toolbox.dev/llms.txt), read 10 Nov 2026.) ## Read next - [MCP servers for Looker](/mcp/looker/): 1 listed - [Looker tools in MCP Toolbox for Databases and Tableau MCP, side by side](/compare/looker-mcp-toolbox-vs-tableau-mcp/) - [The Context Weight Index](/context-weight/) ## Sources - raw.githubusercontent.com/googleapis/genai-toolbox/main/README.md: https://raw.githubusercontent.com/googleapis/genai-toolbox/main/README.md (read 10 Nov 2026) - github.com/googleapis/mcp-toolbox/releases: https://github.com/googleapis/mcp-toolbox/releases (read 10 Nov 2026) - mcp-toolbox.dev/integrations/looker/tools/looker-get-explore: https://mcp-toolbox.dev/integrations/looker/tools/looker-get-explore/ (read 10 Nov 2026) - mcp-toolbox.dev/llms.txt: https://mcp-toolbox.dev/llms.txt (read 10 Nov 2026) - mcp-toolbox.dev/integrations/looker/source: https://mcp-toolbox.dev/integrations/looker/source/ (read 10 Nov 2026) - mcp-toolbox.dev/integrations/looker/tools/looker-health-analyze: https://mcp-toolbox.dev/integrations/looker/tools/looker-health-analyze/ (read 10 Nov 2026) - github.com/googleapis/mcp-toolbox: https://github.com/googleapis/mcp-toolbox (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/servers/looker-mcp-toolbox/ --- # Monte Carlo MCP Server: 15 tools, remote (hosted), from Monte Carlo SERVER, read 10 Nov 2026 **The short answer.** Monte Carlo MCP Server is an official MCP server for Monte Carlo, published by Monte Carlo and read from its own source on 10 Nov 2026. It runs remote (hosted), and 15 tools are named. Version read: Hosted at mcp.getmontecarlo.com/mcp, no version published. Last release: none found on 10 Nov 2026. - **Data tool:** [Monte Carlo](/mcp/monte-carlo/) - **Publisher:** Monte Carlo - **Kind:** Official - **Repository or docs:** [docs.getmontecarlo.com/docs/mcp-server](https://docs.getmontecarlo.com/docs/mcp-server) - **Version or commit:** Hosted at mcp.getmontecarlo.com/mcp, no version published - **Last release:** none found on 10 Nov 2026 - **Transport:** Remote (hosted) - **Auth:** OAuth 2.1 with Dynamic Client Registration (user identity, Editor role or above), or MCP-server-type API key id:secret as Bearer token via mcp-remote. - **Access:** Configurable - **Tools listed:** 15 named in the evidence - **Tool definitions read at:** none found on 10 Nov 2026 - **Context weight:** Not measured: no published tool-definition text was read on 10 Nov 2026 - **Status:** Exists on 10 Nov 2026 as a hosted first-party server; tool names are partial (those named in the docs), full tool definitions were not published as text, so no schema file was saved. - **Read:** 10 Nov 2026 ## In one sentence Monte Carlo's hosted observability server uses per-user OAuth or scoped keys, a per-request read-only header and tool allow-lists. ## What Monte Carlo MCP Server exposes 15 tools are named in the evidence read on 10 Nov 2026. `search`, `get_table`, `get_asset_lineage`, `get_alerts`, `alert_assessment`, `update_alert`, `get_monitors`, `create_or_update_table_monitor`, `get_use_cases`, `get_use_case_tables`, `get_use_case_table_summary`, `get_agent_metadata`, `get_agent_conversation`, `get_agent_trace`, `create_agent_metric_monitor` ## Not found in the sources read - Documentation that get_asset_lineage returns column-level lineage, transformation SQL, BI fields or metrics; the MCP pages name only asset lineage. [Source](https://docs.getmontecarlo.com/docs/mcp-server). - Glossary terms, definitions or classification tools in the Monte Carlo MCP server docs; the tools are alerts, monitors, assets, lineage and data operations. [Source](https://docs.getmontecarlo.com/docs/mcp-server-reference-and-operations). - A list of warehouses and BI systems on the MCP pages (the server fronts the Monte Carlo account's monitored assets). [Source](https://docs.getmontecarlo.com/docs/mcp-server). - A freshness statement on the MCP pages (the server reads live alert and monitor state from the Monte Carlo account). [Source](https://docs.getmontecarlo.com/docs/mcp-server). ## The evidence on Monte Carlo MCP Server, by topic ### Publisher and releases - The server is first-party, run by Monte Carlo at mcp.getmontecarlo.com/mcp, not distributed as a package. (Monte Carlo's own claim; [source](https://docs.getmontecarlo.com/docs/ai-security), read 10 Nov 2026.) - No version number or public release date is published for the hosted server; the docs page carries an updatedAt of 2 Nov 2026. (observed in the source; [source](https://docs.getmontecarlo.com/docs/mcp-server), read 10 Nov 2026.) ### Clients, auth and scoping - Monte Carlo says connected agents can investigate alerts, explore assets and lineage, and create and manage monitors. (Monte Carlo's own claim; [source](https://docs.getmontecarlo.com/docs/mcp-server), read 10 Nov 2026.) - OAuth lets any HTTP-transport MCP client connect with no API keys, and the connector is listed in the Claude Connectors Directory. (Monte Carlo's own claim; [source](https://docs.getmontecarlo.com/docs/mcp-server), read 10 Nov 2026.) - An Editor role or above is required to use the server. (Monte Carlo's own claim; [source](https://docs.getmontecarlo.com/docs/mcp-server), read 10 Nov 2026.) - An agent_observability toolset adds get_agent_metadata, get_agent_conversation and get_agent_trace for inspecting AI agents. (Monte Carlo's own claim; [source](https://docs.getmontecarlo.com/docs/mcp-server-reference-and-operations), read 10 Nov 2026.) ### Access and safety - An x-mcp-readonly header hides write tools such as update_alert and create_or_update_* monitors and blocks calls to them. (Monte Carlo's own claim; [source](https://docs.getmontecarlo.com/docs/mcp-server-reference-and-operations), read 10 Nov 2026.) - Per-request x-mcp-tools allow-lists narrow tools and are enforced on call, not only on listing. (Monte Carlo's own claim; [source](https://docs.getmontecarlo.com/docs/mcp-server-reference-and-operations), read 10 Nov 2026.) - API-key access uses MCP-server-specific scoped keys; standard API keys do not work. (Monte Carlo's own claim; [source](https://docs.getmontecarlo.com/docs/mcp-server), read 10 Nov 2026.) - The MCP server can be restricted by IP using Network Access Control Lists. (Monte Carlo's own claim; [source](https://docs.getmontecarlo.com/docs/mcp-server-reference-and-operations), read 10 Nov 2026.) - External security audits and penetration tests are done at least annually, covering MCP surfaces. (Monte Carlo's own claim; [source](https://docs.getmontecarlo.com/docs/ai-security), read 10 Nov 2026.) ### Lineage and impact - Monte Carlo's MCP server includes a get_asset_lineage tool, used to see how data flows into a table. (Monte Carlo's own claim; [source](https://docs.getmontecarlo.com/docs/mcp-server), read 10 Nov 2026.) ### Definitions, ownership and rules - Governance-like context comes from use cases and critical-table groupings (get_use_cases, get_use_case_tables) in the extended toolset. (Monte Carlo's own claim; [source](https://docs.getmontecarlo.com/docs/mcp-server-reference-and-operations), read 10 Nov 2026.) ### Profile - Monte Carlo also publishes an Agent Toolkit (mc-agent-toolkit) of skills for monitoring and code-change validation, separate from the server. (Monte Carlo's own claim; [source](https://docs.getmontecarlo.com/docs/mcp-server), read 10 Nov 2026.) ## Read next - [MCP servers for Monte Carlo](/mcp/monte-carlo/): 1 listed - [Monte Carlo MCP Server and SYNQ MCP (Coalesce Quality), side by side](/compare/monte-carlo-mcp-vs-synq-mcp/) - [Monte Carlo MCP Server and Sifflet MCP Server, side by side](/compare/monte-carlo-mcp-vs-sifflet-mcp/) - [The Context Weight Index](/context-weight/) ## Sources - docs.getmontecarlo.com/docs/mcp-server: https://docs.getmontecarlo.com/docs/mcp-server (read 10 Nov 2026) - docs.getmontecarlo.com/docs/mcp-server-reference-and-operations: https://docs.getmontecarlo.com/docs/mcp-server-reference-and-operations (read 10 Nov 2026) - docs.getmontecarlo.com/docs/ai-security: https://docs.getmontecarlo.com/docs/ai-security (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/servers/monte-carlo-mcp/ --- # Omni MCP Server: 26 tools, remote (hosted), from Omni Analytics SERVER, read 10 Nov 2026 **The short answer.** Omni MCP Server is an official MCP server for Omni, published by Omni Analytics and read from its own source on 10 Nov 2026. It runs remote (hosted), and 26 tools are named. Version read: Hosted service (no versioned release). Last release: none found on 10 Nov 2026. - **Data tool:** [Omni](/mcp/omni/) - **Publisher:** Omni Analytics - **Kind:** Official - **Repository or docs:** [docs.omni.co/ai/mcp](https://docs.omni.co/ai/mcp) - **Version or commit:** Hosted service (no versioned release) - **Last release:** none found on 10 Nov 2026 - **Transport:** Remote (hosted) - **Auth:** OAuth 2.1 with PKCE (creates an MCP OAuth PAT for the user), or an API key (personal token or Organization API key; Organization keys can act as a user with X-MCP-User-ID). - **Access:** Configurable - **Tools listed:** 26 named in the evidence - **Tool definitions read at:** [docs.omni.co/ai/mcp/tools.md](https://docs.omni.co/ai/mcp/tools.md) - **Context weight:** 466 tokens for 27 tool definitions (27 with a description), measured from the docs page read - **Status:** Exists on 10 Nov 2026 as Omni's hosted MCP server (OAuth callback via callbacks.omniapp.co); admins toggle tool capabilities in AI Hub > MCP. The docs page lists 25 tools. - **Read:** 10 Nov 2026 ## In one sentence Omni's hosted server runs as a specific Omni user and exposes models, topics, saved queries and apps, so answers follow the semantic model and the user's permissions. ## What Omni MCP Server exposes 26 tools are named in the evidence read on 10 Nov 2026; the tool-definition source measured for the Context Weight Index lists 27. `pickModel`, `pickTopic`, `getData`, `runQuery`, `listLlms`, `askOmni`, `checkStatus`, `cancelJob`, `listBranches`, `listSkills`, `submitFeedback`, `searchContent`, `listDocumentQueries`, `runSavedQuery`, `startDashboardDownload`, `checkDashboardDownload`, `fetchDashboardDownload`, `getApp`, `getAppDraft`, `putApp`, `patchApp`, `publishDraft`, `showApp`, `import-claude-dashboard-from-url`, `get-dashboard-import-job-status`, `queryUsageAnalytics` ## Not found in the sources read - A lineage tool among the Omni MCP tools (no column or upstream-table lineage tool documented). [Source](https://docs.omni.co/ai/mcp/tools.md). - Scheduled or event-driven metadata collection in the Omni MCP docs; tools run queries and read content at call time. [Source](https://docs.omni.co/ai/mcp/tools.md). ## The evidence on Omni MCP Server, by topic ### Publisher and releases - Omni documents its own MCP server for querying Omni data and searching Omni docs from external AI tools. (Omni Analytics's own claim; [source](https://docs.omni.co/ai/mcp/tools.md), read 10 Nov 2026.) - No versioned release is published for the hosted server; it is enabled by an organization setting (AI Hub > MCP, enabled by default). (Omni Analytics's own claim; [source](https://docs.omni.co/ai/mcp/tools.md), read 10 Nov 2026.) ### Clients, auth and scoping - MCP tools run with the permissions of a specific Omni user, so results include only what that user can access. (Omni Analytics's own claim; [source](https://docs.omni.co/ai/mcp/tools.md), read 10 Nov 2026.) - With OAuth, all MCP queries use the user's Omni permissions and access controls. (Omni Analytics's own claim; [source](https://docs.omni.co/ai/mcp/authentication.md), read 10 Nov 2026.) - An Organization API key can run tools as another user via the X-MCP-User-ID header, and X-MCP-Model-ID can pin a model. (Omni Analytics's own claim; [source](https://docs.omni.co/ai/mcp/tools.md), read 10 Nov 2026.) ### Access and safety - The tool list includes app write tools (putApp, patchApp, publishDraft) so the server is not read-only; edits land on a shared draft until publishDraft runs. (Omni Analytics's own claim; [source](https://docs.omni.co/ai/mcp/tools.md), read 10 Nov 2026.) - Organization Admins can individually enable or disable tool capabilities on the MCP settings page. (Omni Analytics's own claim; [source](https://docs.omni.co/ai/mcp/tools.md), read 10 Nov 2026.) - A user-scoped key lists a tool only if the member holds the permission it needs. (Omni Analytics's own claim; [source](https://docs.omni.co/ai/mcp/tools.md), read 10 Nov 2026.) - QueryUsageAnalytics uses read-only queries on usage data and is limited to Organization Admins. (Omni Analytics's own claim; [source](https://docs.omni.co/ai/mcp/tools.md), read 10 Nov 2026.) ### What it reaches - The same MCP server can serve embed users, scoped to each user's permissions. (Omni Analytics's own claim; [source](https://docs.omni.co/llms.txt), read 10 Nov 2026.) - Setup pages exist for ChatGPT, Claude Code, Claude Desktop, Codex, Cursor, Microsoft Copilot and VS Code. (Omni Analytics's own claim; [source](https://docs.omni.co/sitemap.xml), read 10 Nov 2026.) ### Lineage and impact - PickModel returns the models available to the calling user and pickTopic returns the topics in the chosen model and selects the most relevant one. (Omni Analytics's own claim; [source](https://docs.omni.co/ai/mcp/tools.md), read 10 Nov 2026.) - SearchContent finds existing dashboards and apps by name, folder or tile name, and listDocumentQueries lists the saved queries in a document. (Omni Analytics's own claim; [source](https://docs.omni.co/ai/mcp/tools.md), read 10 Nov 2026.) ### Definitions, ownership and rules - Queries run against Omni's modeled topics, and the Omni docs publish guidance on configuring AI context at model, topic and view level. (Omni Analytics's own claim; [source](https://docs.omni.co/ai/mcp/tools.md), read 10 Nov 2026.) - ListSkills finds a saved Omni skill by name, identifier or description and reads its instructions without running it. (Omni Analytics's own claim; [source](https://docs.omni.co/ai/mcp/tools.md), read 10 Nov 2026.) ### Freshness and upkeep - GetData and runQuery execute queries on request, and checkStatus and cancelJob manage running jobs. (Omni Analytics's own claim; [source](https://docs.omni.co/ai/mcp/tools.md), read 10 Nov 2026.) ### Profile - The OAuth connection URL routes requests to the user's Omni organization through callbacks.omniapp.co/callback/mcp. (Omni Analytics's own claim; [source](https://docs.omni.co/ai/mcp/authentication.md), read 10 Nov 2026.) ## Read next - [MCP servers for Omni](/mcp/omni/): 1 listed - [Hex MCP server and Omni MCP Server, side by side](/compare/hex-mcp-vs-omni-mcp/) - [The Context Weight Index](/context-weight/) ## Sources - docs.omni.co/ai/mcp/tools.md: https://docs.omni.co/ai/mcp/tools.md (read 10 Nov 2026) - docs.omni.co/ai/mcp/authentication.md: https://docs.omni.co/ai/mcp/authentication.md (read 10 Nov 2026) - docs.omni.co/llms.txt: https://docs.omni.co/llms.txt (read 10 Nov 2026) - docs.omni.co/sitemap.xml: https://docs.omni.co/sitemap.xml (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/servers/omni-mcp/ --- # OpenMetadata MCP Server: 15 tools, remote (hosted), from OpenMetadata SERVER, read 10 Nov 2026 **The short answer.** OpenMetadata MCP Server is an official MCP server for OpenMetadata, published by OpenMetadata and read from its own source on 10 Nov 2026. It runs remote (hosted), and 15 tools are named. Version read: 2.0.7 (bundled with platform release 2.0.7-release; 2.1.0-rc1 also listed). Last release: 27 Oct 2026. - **Data tool:** [OpenMetadata](/mcp/openmetadata/) - **Publisher:** OpenMetadata - **Kind:** Official - **Repository or docs:** [docs.open-metadata.org/v2.0.x/how-to-guides/mcp.md](https://docs.open-metadata.org/v2.0.x/how-to-guides/mcp.md) - **Version or commit:** 2.0.7 (bundled with platform release 2.0.7-release; 2.1.0-rc1 also listed) - **Last release:** 27 Oct 2026 - **Transport:** Remote (hosted) - **Auth:** OAuth 2.0 via existing SSO (per user), personal access token, or Bot JWT as Authorization Bearer. - **Access:** Read and write - **Tools listed:** 15 named in the evidence - **Tool definitions read at:** [docs.open-metadata.org/v2.0.x/how-to-guides/mcp/reference.md](https://docs.open-metadata.org/v2.0.x/how-to-guides/mcp/reference.md) - **Context weight:** 1,249 tokens for 15 tool definitions (15 with a description), measured from the docs page read - **Status:** Exists on 10 Nov 2026; built into the OpenMetadata server (not a separate package). Tool access is governed by the caller's role and policies; no read-only switch is documented on the pages read. - **Read:** 10 Nov 2026 ## In one sentence OpenMetadata's built-in server offers per-user OAuth, persona-scoped context, table and optional column lineage, and write tools for lineage, entities and tests. ## What OpenMetadata MCP Server exposes 15 tools are named in the evidence read on 10 Nov 2026. `search_metadata`, `semantic_search`, `company_context`, `get_entity_details`, `get_user_context`, `get_persona_context`, `find_context`, `get_entity_lineage`, `create_lineage`, `root_cause_analysis`, `create_entity`, `describe_entity_type`, `patch_entity`, `get_test_definitions`, `create_test_case` ## Not found in the sources read - Documentation that the MCP lineage tools return transformation SQL or logic. [Source](https://docs.open-metadata.org/v2.0.x/how-to-guides/mcp/reference.md). - Freshness or collection-cadence statement on the OpenMetadata MCP pages. [Source](https://docs.open-metadata.org/v2.0.x/how-to-guides/mcp.md). ## The evidence on OpenMetadata MCP Server, by topic ### Publisher and releases - The MCP server is built into OpenMetadata and installed and enabled by default. (OpenMetadata's own claim; [source](https://docs.open-metadata.org/v2.0.x/how-to-guides/mcp.md), read 10 Nov 2026.) - OpenMetadata releases 2.0.7 on 27 Oct 2026 (GitHub release page); the MCP server version follows the platform release. (observed in the source; [source](https://github.com/open-metadata/OpenMetadata/releases/tag/2.0.7-release), read 10 Nov 2026.) - The release timestamp is 27 Oct 2026. (observed in the source; [source](https://github.com/open-metadata/OpenMetadata/releases/tag/2.0.7-release), read 10 Nov 2026.) ### Clients, auth and scoping - Get_user_context returns the authenticated caller's identity, roles, teams, domains and active persona, with no parameter to look up another user. (OpenMetadata's own claim; [source](https://docs.open-metadata.org/v2.0.x/how-to-guides/mcp/reference.md), read 10 Nov 2026.) - A personal access token gives the agent the same role and access policy as the user. (OpenMetadata's own claim; [source](https://docs.open-metadata.org/v2.0.x/how-to-guides/mcp.md), read 10 Nov 2026.) - OAuth 2.0 runs tool calls with the user's existing SSO login (Google, Azure, Okta, Auth0, LDAP, SAML). (OpenMetadata's own claim; [source](https://docs.open-metadata.org/v2.0.x/how-to-guides/mcp.md), read 10 Nov 2026.) ### Access and safety - Every tool is labeled Read or Write in the reference. (OpenMetadata's own claim; [source](https://docs.open-metadata.org/v2.0.x/how-to-guides/mcp/reference.md), read 10 Nov 2026.) - Unattended agents should use a Bot service account with a role-based access policy, kept separate from personal logins and auditable. (OpenMetadata's own claim; [source](https://docs.open-metadata.org/v2.0.x/how-to-guides/mcp.md), read 10 Nov 2026.) - The transport is Streamable HTTP (MCP spec 26 Mar 2025). (OpenMetadata's own claim; [source](https://docs.open-metadata.org/v2.0.x/how-to-guides/mcp/connect.md), read 10 Nov 2026.) ### What it reaches - The reference lists asset types: database, dashboard, pipeline, topic, mlmodel, container, searchIndex, storedProcedure and API services; no named vendor list is on this page. (OpenMetadata's own claim; [source](https://docs.open-metadata.org/v2.0.x/how-to-guides/mcp/reference.md), read 10 Nov 2026.) ### Lineage and impact - Get_entity_lineage returns upstream and downstream lineage for any entity, with an includeColumnLineage option (default false, table level only), up to 10 hops. (OpenMetadata's own claim; [source](https://docs.open-metadata.org/v2.0.x/how-to-guides/mcp/reference.md), read 10 Nov 2026.) - Create_lineage lets an agent write lineage edges, and root_cause_analysis is a lineage-and-impact tool. (OpenMetadata's own claim; [source](https://docs.open-metadata.org/v2.0.x/how-to-guides/mcp/reference.md), read 10 Nov 2026.) ### Definitions, ownership and rules - Search_metadata filters on owners, tags, tiers and glossary terms and entity types include glossary, glossaryTerm, metric, domain and dataProduct. (OpenMetadata's own claim; [source](https://docs.open-metadata.org/v2.0.x/how-to-guides/mcp/reference.md), read 10 Nov 2026.) - Find_context and get_persona_context serve shared AI context documents curated per persona. (OpenMetadata's own claim; [source](https://docs.open-metadata.org/v2.0.x/how-to-guides/mcp/reference.md), read 10 Nov 2026.) - Patch_entity and create_entity write metadata (create never modifies an existing entity; patch handles edits). (OpenMetadata's own claim; [source](https://docs.open-metadata.org/v2.0.x/how-to-guides/mcp/reference.md), read 10 Nov 2026.) ### Profile - OpenMetadata's MCP server is described as embedded in an open knowledge graph of metadata, with supported SSO providers including Google, Azure, Okta, Auth0, LDAP and SAML. (OpenMetadata's own claim; [source](https://docs.open-metadata.org/v2.0.x/how-to-guides/mcp.md), read 10 Nov 2026.) ## Read next - [MCP servers for OpenMetadata](/mcp/openmetadata/): 1 listed - [DataHub MCP Server and OpenMetadata MCP Server, side by side](/compare/datahub-mcp-vs-openmetadata-mcp/) - [The Context Weight Index](/context-weight/) ## Sources - docs.open-metadata.org/v2.0.x/how-to-guides/mcp/reference.md: https://docs.open-metadata.org/v2.0.x/how-to-guides/mcp/reference.md (read 10 Nov 2026) - docs.open-metadata.org/v2.0.x/how-to-guides/mcp.md: https://docs.open-metadata.org/v2.0.x/how-to-guides/mcp.md (read 10 Nov 2026) - github.com/open-metadata/OpenMetadata/releases/tag/2.0.7-release: https://github.com/open-metadata/OpenMetadata/releases/tag/2.0.7-release (read 10 Nov 2026) - docs.open-metadata.org/v2.0.x/how-to-guides/mcp/connect.md: https://docs.open-metadata.org/v2.0.x/how-to-guides/mcp/connect.md (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/servers/openmetadata-mcp/ --- # Power BI Modeling MCP Server: 21 tools, remote and local, from Microsoft SERVER, read 10 Nov 2026 **The short answer.** Power BI Modeling MCP Server is an official MCP server for Power BI, published by Microsoft and read from its own source on 10 Nov 2026. It runs remote and local, and 21 tools are named. Version read: 1.0.1. Last release: 6 Nov 2026. - **Data tool:** [Power BI](/mcp/power-bi/) - **Publisher:** Microsoft - **Kind:** Official - **Repository or docs:** [github.com/microsoft/powerbi-modeling-mcp](https://github.com/microsoft/powerbi-modeling-mcp) - **Version or commit:** 1.0.1 - **Last release:** 6 Nov 2026 - **Transport:** Remote and local - **Auth:** Local: Microsoft Entra interactive sign-in (default) or service principal (client secret or certificate). Hosted: Entra ID as the signed-in user. - **Access:** Configurable - **Tools listed:** 21 named in the evidence - **Tool definitions read at:** [raw.githubusercontent.com/microsoft/powerbi-modeling-mcp/main/README.md](https://raw.githubusercontent.com/microsoft/powerbi-modeling-mcp/main/README.md) - **Context weight:** 387 tokens for 21 tool definitions (21 with a description), measured from the README read - **Status:** Exists on 10 Nov 2026. Microsoft documents two Power BI MCP servers: the Authoring server (local GA over stdio from microsoft/powerbi-modeling-mcp, hosted preview at api.fabric.microsoft.com/v1/mcp/powerbi/authoring and Fabric IQ for answering questions over governed data; an earlier consumption MCP remains documented. Hosted authoring tool names are not published, so the tool list is the local server's. - **Read:** 10 Nov 2026 ## In one sentence Microsoft's authoring server reads and writes semantic-model objects (tables, columns, DAX measures, relationships, RLS roles, TMDL) under the user's Fabric permissions, with a read-only switch. ## What Power BI Modeling MCP Server exposes 21 tools are named in the evidence read on 10 Nov 2026. `connection_operations`, `database_operations`, `transaction_operations`, `model_operations`, `table_operations`, `column_operations`, `measure_operations`, `relationship_operations`, `dax_query_operations`, `trace_operations`, `partition_operations`, `user_hierarchy_operations`, `calculation_group_operations`, `security_role_operations`, `perspective_operations`, `named_expression_operations`, `function_operations`, `culture_operations`, `object_translation_operations`, `calendar_operations`, `query_group_operations` ## Not found in the sources read - A lineage tool in the Power BI Authoring MCP tool table (no upstream-table, column-lineage or report-dependency tool). [Source](https://raw.githubusercontent.com/microsoft/powerbi-modeling-mcp/main/README.md). - Scheduled metadata collection or lineage refresh in the Power BI Authoring MCP README or Learn page. [Source](https://raw.githubusercontent.com/microsoft/powerbi-modeling-mcp/main/README.md). ## The evidence on Power BI Modeling MCP Server, by topic ### Publisher and releases - Microsoft publishes the Power BI Authoring MCP Server in its own microsoft GitHub organization and documents it on Microsoft Learn. (Microsoft's own claim; [source](https://learn.microsoft.com/en-us/power-bi/developer/mcp/power-bi-authoring-mcp), read 10 Nov 2026.) - The latest GitHub release is v1.0.1 (previous v1.0.0, then beta tags), timestamped 6 Nov 2026. (observed in the source; [source](https://github.com/microsoft/powerbi-modeling-mcp/releases), read 10 Nov 2026.) - The npm package @microsoft/powerbi-modeling-mcp has latest 1.0.1 published 6 Nov 2026, and 1.0.0 on 25 Sep 2026. (observed in the source; [source](https://registry.npmjs.org/@microsoft/powerbi-modeling-mcp), read 10 Nov 2026.) - The local server is generally available and the hosted endpoint is in preview. (Microsoft's own claim; [source](https://learn.microsoft.com/en-us/power-bi/developer/mcp/power-bi-authoring-mcp), read 10 Nov 2026.) ### Clients, auth and scoping - The hosted server prompts for Microsoft Entra sign-in on first tool call and then acts with the user's permissions. (Microsoft's own claim; [source](https://learn.microsoft.com/en-us/power-bi/developer/mcp/power-bi-authoring-mcp), read 10 Nov 2026.) - Fabric IQ, Microsoft's consumption server, answers questions over semantic models and respects permissions, row-level security and object-level security. (Microsoft's own claim; [source](https://learn.microsoft.com/en-us/power-bi/developer/mcp/mcp-servers-overview), read 10 Nov 2026.) - Microsoft states that clients operate under the user's Fabric RBAC permissions. (Microsoft's own claim; [source](https://raw.githubusercontent.com/microsoft/powerbi-modeling-mcp/main/README.md), read 10 Nov 2026.) ### Access and safety - The --readonly option is described as safe mode that prevents any write operations to the semantic model; --readwrite is the default with a confirmation prompt. (Microsoft's own claim; [source](https://raw.githubusercontent.com/microsoft/powerbi-modeling-mcp/main/README.md), read 10 Nov 2026.) - Blocking access requires disabling the XMLA endpoint because no tenant admin setting specifically blocks this server. (Microsoft's own claim; [source](https://raw.githubusercontent.com/microsoft/powerbi-modeling-mcp/main/README.md), read 10 Nov 2026.) - With only Build permission the server can only run DAX queries; Write permission is needed to change model objects. (Microsoft's own claim; [source](https://learn.microsoft.com/en-us/power-bi/developer/mcp/power-bi-authoring-mcp), read 10 Nov 2026.) ### What it reaches - The local server connects to semantic models in Power BI Desktop, Fabric workspaces and Power BI Project (PBIP) files; --compatibility Full also targets Analysis Services databases. (Microsoft's own claim; [source](https://raw.githubusercontent.com/microsoft/powerbi-modeling-mcp/main/README.md), read 10 Nov 2026.) - Hosted uses Streamable HTTP and Entra ID; local uses stdio, with Desktop, PBIP files, transactions and traces only on local. (Microsoft's own claim; [source](https://learn.microsoft.com/en-us/power-bi/developer/mcp/power-bi-authoring-mcp), read 10 Nov 2026.) ### Lineage and impact - The server exposes model-object tools for tables, columns, measures, relationships, partitions, hierarchies, calculation groups and named expressions (model definition level). (Microsoft's own claim; [source](https://raw.githubusercontent.com/microsoft/powerbi-modeling-mcp/main/README.md), read 10 Nov 2026.) - Database_operations can import and export TMDL folders and deploy to Fabric, giving agents the model definition as code. (Microsoft's own claim; [source](https://raw.githubusercontent.com/microsoft/powerbi-modeling-mcp/main/README.md), read 10 Nov 2026.) ### Definitions, ownership and rules - Tools manage security roles with row-level security, perspectives, cultures and object translations, which are governance and meaning objects inside the model. (Microsoft's own claim; [source](https://raw.githubusercontent.com/microsoft/powerbi-modeling-mcp/main/README.md), read 10 Nov 2026.) - The server can evaluate a model against a set of modeling practice rules and generate documentation for an existing model. (Microsoft's own claim; [source](https://learn.microsoft.com/en-us/power-bi/developer/mcp/mcp-servers-overview), read 10 Nov 2026.) ### Freshness and upkeep - The server connects to semantic models through the XMLA endpoint, so it reads and writes the live model rather than a collected copy. (Microsoft's own claim; [source](https://raw.githubusercontent.com/microsoft/powerbi-modeling-mcp/main/README.md), read 10 Nov 2026.) ### Profile - Use of the software is subject to the Power BI Authoring MCP Server EULA, and NPX installs require explicit acceptance. (Microsoft's own claim; [source](https://raw.githubusercontent.com/microsoft/powerbi-modeling-mcp/main/README.md), read 10 Nov 2026.) - Microsoft advises the hosted server for Fabric workspaces and not registering both local and remote at once. (Microsoft's own claim; [source](https://learn.microsoft.com/en-us/power-bi/developer/mcp/power-bi-authoring-mcp), read 10 Nov 2026.) ## Read next - [MCP servers for Power BI](/mcp/power-bi/): 1 listed - [Power BI Modeling MCP Server and Tableau MCP, side by side](/compare/powerbi-modeling-mcp-vs-tableau-mcp/) - [The Context Weight Index](/context-weight/) ## Sources - learn.microsoft.com/en-us/power-bi/developer/mcp/power-bi-authoring-mcp: https://learn.microsoft.com/en-us/power-bi/developer/mcp/power-bi-authoring-mcp (read 10 Nov 2026) - github.com/microsoft/powerbi-modeling-mcp/releases: https://github.com/microsoft/powerbi-modeling-mcp/releases (read 10 Nov 2026) - registry.npmjs.org/@microsoft/powerbi-modeling-mcp: https://registry.npmjs.org/@microsoft/powerbi-modeling-mcp (read 10 Nov 2026) - raw.githubusercontent.com/microsoft/powerbi-modeling-mcp/main/README.md: https://raw.githubusercontent.com/microsoft/powerbi-modeling-mcp/main/README.md (read 10 Nov 2026) - learn.microsoft.com/en-us/power-bi/developer/mcp/mcp-servers-overview: https://learn.microsoft.com/en-us/power-bi/developer/mcp/mcp-servers-overview (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/servers/powerbi-modeling-mcp/ --- # Secoda MCP: 8 tools, remote and local, from Secoda SERVER, read 10 Nov 2026 **The short answer.** Secoda MCP is an official MCP server for Secoda, published by Secoda and read from its own source on 10 Nov 2026. It runs remote and local, and 8 tools are named. Version read: No releases; main branch last commit 26 Jan 2026. Last release: none found on 10 Nov 2026. - **Data tool:** [Secoda](/mcp/secoda/) - **Publisher:** Secoda - **Kind:** Official - **Repository or docs:** [github.com/secoda/secoda-mcp](https://github.com/secoda/secoda-mcp) - **Version or commit:** No releases; main branch last commit 26 Jan 2026 - **Last release:** none found on 10 Nov 2026 - **Transport:** Remote and local - **Auth:** Secoda API token (API_TOKEN env var) for the local FastMCP wrapper; docs also describe a Secoda-hosted MCP endpoint authenticated against workspace and AI permissions. - **Access:** Not stated in the source read - **Tools listed:** 8 named in the evidence - **Tool definitions read at:** [github.com/secoda/secoda-mcp/blob/main/server.py](https://github.com/secoda/secoda-mcp/blob/main/server.py) - **Context weight:** 63 tokens for 6 tool definitions (6 with a description), measured from the source code read - **Status:** Exists on 10 Nov 2026: open-source wrapper with 10 commits and no GitHub releases, plus a Secoda-hosted MCP described in docs.secoda.co (eight tools listed there; the wrapper code implements run_sql, search_data_assets, search_documentation, retrieve_entity, entity_lineage and glossary). - **Read:** 10 Nov 2026 ## In one sentence Secoda's server combines catalog search, glossary, lineage and SQL over a workspace, as a hosted endpoint and an open-source wrapper. ## What Secoda MCP exposes 8 tools are named in the evidence read on 10 Nov 2026; the tool-definition source measured for the Context Weight Index lists 6. `search_data_assets`, `search`, `run_sql`, `retrieve_entity`, `entity_lineage`, `glossary`, `get_secoda_docs`, `chart` ## Not found in the sources read - Documentation of column-level lineage, transformation SQL, BI-field or metric lineage in the Secoda MCP docs. [Source](https://docs.secoda.co/features/ai-assistant/secoda-mcp-server.md). - Ownership, classification or governance-rule propagation along lineage in the Secoda MCP docs. [Source](https://docs.secoda.co/features/ai-assistant/secoda-mcp-server.md). - A list of supported warehouses or BI tools on the Secoda MCP page. [Source](https://docs.secoda.co/features/ai-assistant/secoda-mcp-server.md). - Freshness statement in the Secoda MCP docs. [Source](https://docs.secoda.co/features/ai-assistant/secoda-mcp-server.md). ## The evidence on Secoda MCP, by topic ### Publisher and releases - Secoda-mcp is under the secoda GitHub organization and documented in Secoda's own docs; it is MIT licensed Python/FastMCP. (Secoda's own claim; [source](https://docs.secoda.co/features/ai-assistant/secoda-mcp-server.md), read 10 Nov 2026.) - GitHub shows no releases for secoda-mcp, so no version number exists. (observed in the source; [source](https://github.com/secoda/secoda-mcp/releases), read 10 Nov 2026.) - The latest commit on main is a merge of pull request 4 dated 26 Jan 2026. (observed in the source; [source](https://api.github.com/repos/secoda/secoda-mcp/commits?per_page=1), read 10 Nov 2026.) ### Clients, auth and scoping - The Secoda-hosted MCP server lets AI tools search, query SQL, read lineage and glossary terms. (Secoda's own claim; [source](https://docs.secoda.co/features/ai-assistant/secoda-mcp-server.md), read 10 Nov 2026.) - Authentication verifies the user's workspace and AI permissions, and operations respect governance and access controls. (Secoda's own claim; [source](https://docs.secoda.co/features/ai-assistant/secoda-mcp-server.md), read 10 Nov 2026.) - The open-source wrapper authenticates with a single API_TOKEN, i.e. one shared token per server process. (Secoda's own claim; [source](https://docs.secoda.co/features/ai-assistant/secoda-mcp-server.md), read 10 Nov 2026.) ### Access and safety - The tools shown are search, SQL execution, entity retrieval, lineage and glossary; run_sql can execute SQL, and the docs do not describe a read-only mode. (Secoda's own claim; [source](https://docs.secoda.co/features/ai-assistant/secoda-mcp-server.md), read 10 Nov 2026.) - Admin must enable MCP and the user needs the right Secoda permissions. (Secoda's own claim; [source](https://docs.secoda.co/features/ai-assistant/secoda-mcp-server.md), read 10 Nov 2026.) ### Lineage and impact - The entity_lineage tool explores upstream and downstream dependencies for an entity. (Secoda's own claim; [source](https://docs.secoda.co/features/ai-assistant/secoda-mcp-server.md), read 10 Nov 2026.) - In the open-source wrapper entity_lineage takes only an entity_id and no depth or column parameter. (observed in the source; [source](https://github.com/secoda/secoda-mcp/blob/main/server.py), read 10 Nov 2026.) - The run_sql tool executes SQL against connected warehouses. (Secoda's own claim; [source](https://docs.secoda.co/features/ai-assistant/secoda-mcp-server.md), read 10 Nov 2026.) ### Definitions, ownership and rules - A glossary tool returns business terms and definitions. (Secoda's own claim; [source](https://docs.secoda.co/features/ai-assistant/secoda-mcp-server.md), read 10 Nov 2026.) ### Profile - Secoda supports EU, APAC and self-hosted instances by changing the API URL. (Secoda's own claim; [source](https://docs.secoda.co/features/ai-assistant/secoda-mcp-server.md), read 10 Nov 2026.) ## Read next - [MCP servers for Secoda](/mcp/secoda/): 1 listed - [Secoda MCP and Select Star MCP Server, side by side](/compare/secoda-mcp-vs-select-star-mcp/) - [The Context Weight Index](/context-weight/) ## Sources - docs.secoda.co/features/ai-assistant/secoda-mcp-server.md: https://docs.secoda.co/features/ai-assistant/secoda-mcp-server.md (read 10 Nov 2026) - github.com/secoda/secoda-mcp/blob/main/server.py: https://github.com/secoda/secoda-mcp/blob/main/server.py (read 10 Nov 2026) - github.com/secoda/secoda-mcp/releases: https://github.com/secoda/secoda-mcp/releases (read 10 Nov 2026) - api.github.com/repos/secoda/secoda-mcp/commits?per_page=1: https://api.github.com/repos/secoda/secoda-mcp/commits?per_page=1 (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/servers/secoda-mcp/ --- # Select Star MCP Server: 7 tools, remote (hosted), from Select Star SERVER, read 10 Nov 2026 **The short answer.** Select Star MCP Server is an official MCP server for Select Star, published by Select Star and read from its own source on 10 Nov 2026. It runs remote (hosted), and 7 tools are named. Version read: Hosted, no version published. Last release: none found on 10 Nov 2026. - **Data tool:** [Select Star](/mcp/select-star/) - **Publisher:** Select Star (Snowflake agreed to acquire it, announced 24 Nov 2025) - **Kind:** Official - **Repository or docs:** [docs.selectstar.com/features/mcp-server](https://docs.selectstar.com/features/mcp-server) - **Version or commit:** Hosted, no version published - **Last release:** none found on 10 Nov 2026 - **Transport:** Remote (hosted) - **Auth:** Select Star API token sent as Authorization: Bearer header (mcp-remote proxy used for Claude Desktop). - **Access:** Read only - **Tools listed:** 7 named in the evidence - **Tool definitions read at:** [docs.selectstar.com/features/mcp-server](https://docs.selectstar.com/features/mcp-server) - **Context weight:** 98 tokens for 7 tool definitions (7 with a description), measured from the docs page read - **Status:** Exists on 10 Nov 2026 at mcp.production.selectstar.com/mcp. docs.selectstar.com did not resolve from the runner today, so the docs page was read from a web.archive.org snapshot dated 14 Jun 2026; Snowflake announced plans to acquire Select Star on 24 Nov 2025. - **Read:** 10 Nov 2026 ## In one sentence Select Star's hosted, read-only server exposes catalog search, lineage, query history and semantic models with token setup. ## What Select Star MCP Server exposes 7 tools are named in the evidence read on 10 Nov 2026. `search_metadata`, `get_asset_details`, `get_table_info`, `traverse_lineage`, `get_asset_queries`, `search_semantic_models`, `get_semantic_model_details` ## Not found in the sources read - Documentation on the MCP page that traverse_lineage is column-level or returns transformation SQL, BI-field or metric lineage. [Source](https://docs.selectstar.com/features/mcp-server). - A statement that Select Star MCP results are limited to the asking user's own permissions. [Source](https://docs.selectstar.com/features/mcp-server). - Any freshness or collection-cadence statement on the Select Star MCP docs page. [Source](https://docs.selectstar.com/features/mcp-server). ## The evidence on Select Star MCP Server, by topic ### Publisher and releases - The server is hosted by Select Star at mcp.production.selectstar.com/mcp and documented on Select Star's own docs site. (Select Star's own claim; [source](https://docs.selectstar.com/features/mcp-server), read 10 Nov 2026.) - No version number or release date is published for the MCP server; the docs page was last updated 7 months before the 14 Jun 2026 archive snapshot read. (third-party source; [source](https://web.archive.org/web/20260614045618/https://docs.selectstar.com/features/mcp-server), read 10 Nov 2026.) ### Clients, auth and scoping - The MCP server is meant to bring catalog metadata and lineage to Cursor, Claude Desktop and VS Code GitHub Copilot. (Select Star's own claim; [source](https://selectstar.com/product/mcp-for-data), read 10 Nov 2026.) - Authentication is a Select Star API token sent as a Bearer header; no per-user OAuth is described, and the docs do not say results are scoped to the asking user. (Select Star's own claim; [source](https://docs.selectstar.com/features/mcp-server), read 10 Nov 2026.) ### Access and safety - Seven tools are listed, all search/get/traverse operations; no write tools appear on the page. (observed in the source; [source](https://docs.selectstar.com/features/mcp-server), read 10 Nov 2026.) - Troubleshooting says the token must be formatted with a Bearer prefix. (Select Star's own claim; [source](https://docs.selectstar.com/features/mcp-server), read 10 Nov 2026.) ### What it reaches - Select Star says its universal metadata model connects more than 25 integrations (founder post on the Snowflake deal); the MCP page names no systems. (Select Star's own claim; [source](https://selectstar.com/resources/snowflake-acquire-select-star), read 10 Nov 2026.) ### Lineage and impact - The Select Star MCP server has a traverse_lineage tool for lineage relationships and dependencies. (Select Star's own claim; [source](https://docs.selectstar.com/features/mcp-server), read 10 Nov 2026.) - Get_asset_queries returns warehouse queries connected to tables or columns. (Select Star's own claim; [source](https://docs.selectstar.com/features/mcp-server), read 10 Nov 2026.) ### Definitions, ownership and rules - The product page says the MCP server returns descriptions, tags, owners and popularity metrics for agents. (Select Star's own claim; [source](https://selectstar.com/product/mcp-for-data), read 10 Nov 2026.) - It exposes the semantic layer and business glossary, with standardized metrics, dimensions and business terms. (Select Star's own claim; [source](https://selectstar.com/product/mcp-for-data), read 10 Nov 2026.) - Two tools search semantic models and return their full details including YAML. (Select Star's own claim; [source](https://docs.selectstar.com/features/mcp-server), read 10 Nov 2026.) ### Profile - Snowflake announced plans to acquire the Select Star platform (post dated 24 Nov 2025). (Select Star's own claim; [source](https://selectstar.com/resources/snowflake-acquire-select-star), read 10 Nov 2026.) ## Read next - [MCP servers for Select Star](/mcp/select-star/): 1 listed - [Secoda MCP and Select Star MCP Server, side by side](/compare/secoda-mcp-vs-select-star-mcp/) - [The Context Weight Index](/context-weight/) ## Sources - docs.selectstar.com/features/mcp-server: https://docs.selectstar.com/features/mcp-server (read 10 Nov 2026) - selectstar.com/product/mcp-for-data: https://selectstar.com/product/mcp-for-data (read 10 Nov 2026) - selectstar.com/resources/snowflake-acquire-select-star: https://selectstar.com/resources/snowflake-acquire-select-star (read 10 Nov 2026) - web.archive.org/web/20260614045618/https://docs.selectstar.com/features/mcp-server: https://web.archive.org/web/20260614045618/https://docs.selectstar.com/features/mcp-server (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/servers/select-star-mcp/ --- # Sifflet MCP Server: 10 tools, local (stdio), from Sifflet SERVER, read 10 Nov 2026 **The short answer.** Sifflet MCP Server is an official MCP server for Sifflet, published by Sifflet and read from its own source on 10 Nov 2026. It runs local (stdio), and 10 tools are named. Version read: 0.1.7 (PyPI); main at commit 1f390ca. Last release: 3 Aug 2026. - **Data tool:** [Sifflet](/mcp/sifflet/) - **Publisher:** Sifflet - **Kind:** Official - **Repository or docs:** [github.com/siffletdata/sifflet-mcp](https://github.com/siffletdata/sifflet-mcp) - **Version or commit:** 0.1.7 (PyPI); main at commit 1f390ca - **Last release:** 3 Aug 2026 - **Transport:** Local (stdio) - **Auth:** Sifflet API token (Viewer role for most tools, Editor for monitor generation) and backend URL as environment variables. - **Access:** Read and write - **Tools listed:** 10 named in the evidence - **Tool definitions read at:** [raw.githubusercontent.com/siffletdata/sifflet-mcp/1f390ca06d7919214bfe067991539502fdb4b9fc/src/sifflet_mcp/server.py](https://raw.githubusercontent.com/siffletdata/sifflet-mcp/1f390ca06d7919214bfe067991539502fdb4b9fc/src/sifflet_mcp/server.py) - **Context weight:** 662 tokens for 10 tool definitions (10 with a description), measured from the source code read - **Status:** Exists on 10 Nov 2026; Sifflet docs page (updated 1 Oct 2026) points to the siffletdata/sifflet-mcp GitHub repository, MIT license, PyPI 0.1.7 released 3 Aug 2026; listed after the owner asked for one re-check (the docs host docs.sifflet.io did not load on 10 Nov 2026, docs.siffletdata.com did). - **Read:** 10 Nov 2026 ## In one sentence Sifflet's local server covers asset search, incidents, monitors as code and downstream impact from one Sifflet API token. ## What Sifflet MCP Server exposes 10 tools are named in the evidence read on 10 Nov 2026. `asset_by_urn`, `search_asset`, `search_incidents`, `get_incident_by_issue_number`, `get_incident_scope_by_issue_number`, `get_monitor_details_by_id`, `close_incident_by_id`, `open_incident_by_id`, `get_monitor_code_by_description`, `get_downstream_assets_of_asset` ## Not found in the sources read - An audit log or per-call permission prompt described for the MCP server in the README or the docs page. [Source](https://github.com/siffletdata/sifflet-mcp/blob/main/README.md). ## The evidence on Sifflet MCP Server, by topic ### Publisher and releases - The repository is siffletdata/sifflet-mcp under the MIT license; its pyproject names version 0.1.7. (observed in the source; [source](https://raw.githubusercontent.com/siffletdata/sifflet-mcp/main/pyproject.toml), read 10 Nov 2026.) - PyPI lists sifflet-mcp 0.1.7, uploaded 3 Aug 2026; the previous releases are 0.1.6 (3 Sep 2025) and 0.1.5 (2 Jun 2025). (observed in the source; [source](https://pypi.org/project/sifflet-mcp/), read 10 Nov 2026.) - Sifflet's own documentation page for the MCP server points to the GitHub repository for installation. (Sifflet's own claim; [source](https://docs.siffletdata.com/docs/sifflet-mcp-server.md), read 10 Nov 2026.) ### Clients, auth and scoping - The README documents setup for Cursor and Claude Desktop through uvx sifflet-mcp@latest. (observed in the source; [source](https://github.com/siffletdata/sifflet-mcp/blob/main/README.md), read 10 Nov 2026.) ### Access and safety - A Viewer-role token is enough for most tools; get_monitor_code_by_description needs the Editor role on the domain. (Sifflet's own claim; [source](https://github.com/siffletdata/sifflet-mcp/blob/main/README.md), read 10 Nov 2026.) - Two tools change state: close_incident_by_id and open_incident_by_id patch an incident's status. (observed in the source; [source](https://github.com/siffletdata/sifflet-mcp/blob/1f390ca06d7919214bfe067991539502fdb4b9fc/src/sifflet_mcp/server.py), read 10 Nov 2026.) ### What it reaches - The search_asset tool filters by asset type TABLE_AND_VIEW, PIPELINE, DASHBOARD or ML_MODEL. (observed in the source; [source](https://github.com/siffletdata/sifflet-mcp/blob/1f390ca06d7919214bfe067991539502fdb4b9fc/src/sifflet_mcp/server.py), read 10 Nov 2026.) ### Lineage and impact - Sifflet MCP exposes get_downstream_assets_of_asset, which returns all downstream assets of an asset by URN. (observed in the source; [source](https://github.com/siffletdata/sifflet-mcp/blob/1f390ca06d7919214bfe067991539502fdb4b9fc/src/sifflet_mcp/server.py), read 10 Nov 2026.) - Sifflet says the server can start from an incident and trace the downstream assets affected. (Sifflet's own claim; [source](https://github.com/siffletdata/sifflet-mcp/blob/main/README.md), read 10 Nov 2026.) ### Definitions, ownership and rules - Sifflet says asset search returns schema, owners, tags and metadata for tables, views and dashboards. (Sifflet's own claim; [source](https://github.com/siffletdata/sifflet-mcp/blob/main/README.md), read 10 Nov 2026.) ### Freshness and upkeep - The server reads live from the Sifflet API at call time with the configured backend URL; no cached copy is described. (observed in the source; [source](https://github.com/siffletdata/sifflet-mcp/blob/main/README.md), read 10 Nov 2026.) ## Read next - [MCP servers for Sifflet](/mcp/sifflet/): 1 listed - [Monte Carlo MCP Server and Sifflet MCP Server, side by side](/compare/monte-carlo-mcp-vs-sifflet-mcp/) - [The Context Weight Index](/context-weight/) ## Sources - github.com/siffletdata/sifflet-mcp/blob/1f390ca06d7919214bfe067991539502fdb4b9fc/src/sifflet_mcp/server.py: https://github.com/siffletdata/sifflet-mcp/blob/1f390ca06d7919214bfe067991539502fdb4b9fc/src/sifflet_mcp/server.py (read 10 Nov 2026) - github.com/siffletdata/sifflet-mcp/blob/main/README.md: https://github.com/siffletdata/sifflet-mcp/blob/main/README.md (read 10 Nov 2026) - raw.githubusercontent.com/siffletdata/sifflet-mcp/main/pyproject.toml: https://raw.githubusercontent.com/siffletdata/sifflet-mcp/main/pyproject.toml (read 10 Nov 2026) - pypi.org/project/sifflet-mcp: https://pypi.org/project/sifflet-mcp/ (read 10 Nov 2026) - docs.siffletdata.com/docs/sifflet-mcp-server.md: https://docs.siffletdata.com/docs/sifflet-mcp-server.md (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/servers/sifflet-mcp/ --- # Sigma MCP server: tools not named, remote (hosted), from Sigma Computing SERVER, read 10 Nov 2026 **The short answer.** Sigma MCP server is an official MCP server for Sigma, published by Sigma Computing and read from its own source on 10 Nov 2026. It runs remote (hosted), and 0 tools are named. Version read: Hosted service (no versioned release). Last release: none found on 10 Nov 2026. - **Data tool:** [Sigma](/mcp/sigma/) - **Publisher:** Sigma Computing - **Kind:** Official - **Repository or docs:** [help.sigmacomputing.com/docs/use-sigma-mcp-server](https://help.sigmacomputing.com/docs/use-sigma-mcp-server) - **Version or commit:** Hosted service (no versioned release) - **Last release:** none found on 10 Nov 2026 - **Transport:** Remote (hosted) - **Auth:** OAuth (HTTP transport) inheriting the user's Sigma account permissions; API-key access token with scope mcp:access (Beta). - **Access:** Read and write - **Tools listed:** 0 named in the evidence - **Tool definitions read at:** [help.sigmacomputing.com/docs/use-sigma-mcp-server.md](https://help.sigmacomputing.com/docs/use-sigma-mcp-server.md) - **Context weight:** Not measured: The docs page describes features (search, explore, analyze, build workbooks, screenshots, agents) in prose with example prompts but publishes no tool names or per-tool list. - **Status:** Exists on 10 Nov 2026 as a Sigma-hosted MCP server (URL per organization under Profile > Integrations). The docs describe capabilities (search, describe, query, build workbooks, screenshots, call agents) but publish no tool names, so tools is empty. - **Read:** 10 Nov 2026 ## In one sentence Sigma's hosted server searches data models and workbooks, reads column definitions, and queries and builds workbooks under the user's own Sigma permissions. ## What Sigma MCP server exposes No tool names are published in the source read on 10 Nov 2026. No tool names are published in the source read on 10 Nov 2026. ## Not found in the sources read - Lineage or upstream-dependency capability on the Sigma MCP server page. [Source](https://help.sigmacomputing.com/docs/use-sigma-mcp-server). - Scheduled or event-driven metadata collection on the Sigma MCP page; queries run at call time. [Source](https://help.sigmacomputing.com/docs/use-sigma-mcp-server). ## The evidence on Sigma MCP server, by topic ### Publisher and releases - Sigma documents its own Sigma MCP server for searching, describing and querying data and building workbooks. (Sigma Computing's own claim; [source](https://help.sigmacomputing.com/docs/use-sigma-mcp-server), read 10 Nov 2026.) - The server is a hosted endpoint whose URL is copied from Profile > Integrations; no versioned release or package is published. (Sigma Computing's own claim; [source](https://help.sigmacomputing.com/docs/use-sigma-mcp-server), read 10 Nov 2026.) - The API-key access path for the MCP server is labeled Beta. (Sigma Computing's own claim; [source](https://help.sigmacomputing.com/docs/use-sigma-mcp-server), read 10 Nov 2026.) ### Clients, auth and scoping - With OAuth the server inherits the user's Sigma account permissions. (Sigma Computing's own claim; [source](https://help.sigmacomputing.com/docs/use-sigma-mcp-server), read 10 Nov 2026.) - Access needs the 'Use Sigma MCP with OAuth' account-type permission plus at least Can view on documents or Can use on connections. (Sigma Computing's own claim; [source](https://help.sigmacomputing.com/docs/use-sigma-mcp-server), read 10 Nov 2026.) - Agents can call Sigma agents defined in workbooks if the user has Can view on the workbook. (Sigma Computing's own claim; [source](https://help.sigmacomputing.com/docs/use-sigma-mcp-server), read 10 Nov 2026.) ### Access and safety - Workbooks cannot be built when authenticating with an API key, and builds cannot include AI columns or custom SQL elements. (Sigma Computing's own claim; [source](https://help.sigmacomputing.com/docs/use-sigma-mcp-server), read 10 Nov 2026.) - MCP queries can be monitored in the data platform using the "kind":"mcp" query tag. (Sigma Computing's own claim; [source](https://help.sigmacomputing.com/docs/use-sigma-mcp-server), read 10 Nov 2026.) - Orgs with an IP allowlist must allowlist the AI assistant's outbound IPs with scope API. (Sigma Computing's own claim; [source](https://help.sigmacomputing.com/docs/use-sigma-mcp-server), read 10 Nov 2026.) ### What it reaches - Documented clients include Claude, ChatGPT, Codex, Cortex Code and Cursor over HTTP transport. (Sigma Computing's own claim; [source](https://help.sigmacomputing.com/docs/use-sigma-mcp-server), read 10 Nov 2026.) - The configuration table sets transport to HTTP. (Sigma Computing's own claim; [source](https://help.sigmacomputing.com/docs/use-sigma-mcp-server), read 10 Nov 2026.) ### Lineage and impact - Agents can explore data sources and documents by asking about column names, definitions, descriptions, data types and column attributes. (Sigma Computing's own claim; [source](https://help.sigmacomputing.com/docs/use-sigma-mcp-server), read 10 Nov 2026.) - Search covers documents and data sources including data models, warehouse tables, and elements from workbooks and data models. (Sigma Computing's own claim; [source](https://help.sigmacomputing.com/docs/use-sigma-mcp-server), read 10 Nov 2026.) ### Definitions, ownership and rules - The server tells the assistant about existing filters and controls that affect a workbook answer so it can disclose or override them. (Sigma Computing's own claim; [source](https://help.sigmacomputing.com/docs/use-sigma-mcp-server), read 10 Nov 2026.) ### Freshness and upkeep - The server queries only the published version of a document, not custom views, saved views or tagged versions. (Sigma Computing's own claim; [source](https://help.sigmacomputing.com/docs/use-sigma-mcp-server), read 10 Nov 2026.) ### Profile - New workbooks built over MCP are saved in the user's Your documents folder; editing an existing workbook publishes a new version. (Sigma Computing's own claim; [source](https://help.sigmacomputing.com/docs/use-sigma-mcp-server), read 10 Nov 2026.) ## Read next - [MCP servers for Sigma](/mcp/sigma/): 1 listed - [Sigma MCP server and ThoughtSpot MCP Server, side by side](/compare/sigma-mcp-vs-thoughtspot-mcp-server/) - [The Context Weight Index](/context-weight/) ## Sources - help.sigmacomputing.com/docs/use-sigma-mcp-server: https://help.sigmacomputing.com/docs/use-sigma-mcp-server (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/servers/sigma-mcp/ --- # mcp-snowflake-server (community): 8 tools, local (stdio), from isaacwasserman SERVER, read 10 Nov 2026 **The short answer.** mcp-snowflake-server (community) is a community MCP server for Snowflake, published by isaacwasserman and read from its own source on 10 Nov 2026. It runs local (stdio), and 8 tools are named. Version read: 0.4.0. Last release: 10 Apr 2025. Auth: Snowflake user credentials (password, private key, external browser) or connections.toml. - **Data tool:** [Snowflake](/mcp/snowflake/) - **Publisher:** isaacwasserman (community) - **Kind:** Community - **Repository or docs:** [github.com/isaacwasserman/mcp-snowflake-server](https://github.com/isaacwasserman/mcp-snowflake-server) - **Version or commit:** 0.4.0 - **Last release:** 10 Apr 2025 - **Transport:** Local (stdio) - **Auth:** Snowflake user credentials (password, private key, external browser) or connections.toml. - **Access:** Configurable - **Tools listed:** 8 named in the evidence - **Tool definitions read at:** [raw.githubusercontent.com/isaacwasserman/mcp-snowflake-server/main/README.md](https://raw.githubusercontent.com/isaacwasserman/mcp-snowflake-server/main/README.md) - **Context weight:** 103 tokens for 8 tool definitions (8 with a description), measured from the README read - **Status:** Exists as a community server; PyPI 0.4.0 dated 10 Apr 2025, GitHub pushed 7 Oct 2025 (GitHub API). - **Read:** 10 Nov 2026 ## In one sentence A community server for schema browsing and read-only SQL on Snowflake, with write tools off by default. ## What mcp-snowflake-server (community) exposes 8 tools are named in the evidence read on 10 Nov 2026. `read_query`, `write_query`, `create_table`, `list_databases`, `list_schemas`, `list_tables`, `describe_table`, `append_insight` ## Not found in the sources read - A lineage or dependency tool in the README tool list. [Source](https://raw.githubusercontent.com/isaacwasserman/mcp-snowflake-server/main/README.md). ## The evidence on mcp-snowflake-server (community), by topic ### Publisher and releases - The server is a community project by isaacwasserman, not published by Snowflake; license is GPL-3.0 per GitHub API. (isaacwasserman's own claim; [source](https://raw.githubusercontent.com/isaacwasserman/mcp-snowflake-server/main/README.md), read 10 Nov 2026.) - PyPI lists mcp-snowflake-server 0.4.0 uploaded 10 Apr 2025; the GitHub repo was last pushed 7 Oct 2025. (observed in the source; [source](https://pypi.org/pypi/mcp-snowflake-server/json), read 10 Nov 2026.) ### Clients, auth and scoping - The server exposes a memo://insights resource and an append_insight tool that stores agent-discovered insights. (isaacwasserman's own claim; [source](https://raw.githubusercontent.com/isaacwasserman/mcp-snowflake-server/main/README.md), read 10 Nov 2026.) - When prefetch is enabled it exposes per-table schema summaries as resources. (isaacwasserman's own claim; [source](https://raw.githubusercontent.com/isaacwasserman/mcp-snowflake-server/main/README.md), read 10 Nov 2026.) ### Access and safety - Write operations are disabled by default and enabled only with --allow-write. (isaacwasserman's own claim; [source](https://raw.githubusercontent.com/isaacwasserman/mcp-snowflake-server/main/README.md), read 10 Nov 2026.) - Tool exclusion and database/schema/table exclusion patterns are supported. (isaacwasserman's own claim; [source](https://raw.githubusercontent.com/isaacwasserman/mcp-snowflake-server/main/README.md), read 10 Nov 2026.) - Authentication uses Snowflake user credentials, a private key path or external browser; the README shows plain passwords in config. (isaacwasserman's own claim; [source](https://raw.githubusercontent.com/isaacwasserman/mcp-snowflake-server/main/README.md), read 10 Nov 2026.) ### What it reaches - Covers Snowflake only. (isaacwasserman's own claim; [source](https://raw.githubusercontent.com/isaacwasserman/mcp-snowflake-server/main/README.md), read 10 Nov 2026.) ### Lineage and impact - Schema tools list databases, schemas, tables and describe columns; no lineage tool is listed. (isaacwasserman's own claim; [source](https://raw.githubusercontent.com/isaacwasserman/mcp-snowflake-server/main/README.md), read 10 Nov 2026.) ### Definitions, ownership and rules - Describe_table returns column names, types, nullability, defaults and comments, so table and column comments reach the agent. (isaacwasserman's own claim; [source](https://raw.githubusercontent.com/isaacwasserman/mcp-snowflake-server/main/README.md), read 10 Nov 2026.) ### Freshness and upkeep - Metadata is read live from Snowflake on each tool call; no scheduled collection is described. (isaacwasserman's own claim; [source](https://raw.githubusercontent.com/isaacwasserman/mcp-snowflake-server/main/README.md), read 10 Nov 2026.) ### Profile - Run via uvx or uv with stdio for Claude Desktop; install via Smithery is also documented. (isaacwasserman's own claim; [source](https://raw.githubusercontent.com/isaacwasserman/mcp-snowflake-server/main/README.md), read 10 Nov 2026.) ## Read next - [MCP servers for Snowflake](/mcp/snowflake/): 3 listed - [Snowflake Labs MCP Server](/servers/snowflake-labs-mcp/): also for Snowflake - [Snowflake-managed MCP server](/servers/snowflake-managed-mcp/): also for Snowflake - [mcp-snowflake-server (community) and Snowflake-managed MCP server, side by side](/compare/isaacwasserman-mcp-snowflake-server-vs-snowflake-managed-mcp/) - [mcp-snowflake-server (community) and Snowflake Labs MCP Server, side by side](/compare/isaacwasserman-mcp-snowflake-server-vs-snowflake-labs-mcp/) - [The Context Weight Index](/context-weight/) ## Sources - raw.githubusercontent.com/isaacwasserman/mcp-snowflake-server/main/README.md: https://raw.githubusercontent.com/isaacwasserman/mcp-snowflake-server/main/README.md (read 10 Nov 2026) - pypi.org/pypi/mcp-snowflake-server/json: https://pypi.org/pypi/mcp-snowflake-server/json (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/servers/isaacwasserman-mcp-snowflake-server/ --- # Snowflake Labs MCP Server: 6 tools, remote and local, from Snowflake Labs SERVER, read 10 Nov 2026 **The short answer.** Snowflake Labs MCP Server is an official Labs-org MCP server for Snowflake, published by Snowflake Labs and read from its own source on 10 Nov 2026. It runs remote and local, and 6 tools are named. Version read: 1.4.2. Last release: 15 May 2026. - **Data tool:** [Snowflake](/mcp/snowflake/) - **Publisher:** Snowflake Labs - **Kind:** Official, Labs org - **Repository or docs:** [github.com/Snowflake-Labs/mcp](https://github.com/Snowflake-Labs/mcp) - **Version or commit:** 1.4.2 - **Last release:** 15 May 2026 - **Transport:** Remote and local - **Auth:** Snowflake Python connector: password/PAT, key pair, OAuth, SSO, MFA; honors the connection role's RBAC. - **Access:** Configurable - **Tools listed:** 6 named in the evidence - **Tool definitions read at:** [raw.githubusercontent.com/Snowflake-Labs/mcp/main/README.md](https://raw.githubusercontent.com/Snowflake-Labs/mcp/main/README.md) - **Context weight:** Not measured: The Snowflake Labs server, marked deprecated in its README, which describes config-driven tool groups (Cortex Search, Analyst, Agent, object management, SQL execution, semantic views) but publishes no per-tool names or descriptions; facts.json lists these six groups. - **Status:** Exists, but the README states it is deprecated and no longer maintained and points to the Snowflake-managed MCP server (read 10 Nov 2026). - **Read:** 10 Nov 2026 ## In one sentence The Labs server gives config-driven access to Cortex services, object management and semantic views under the connecting role; its README marks it deprecated. ## What Snowflake Labs MCP Server exposes 6 tools are named in the evidence read on 10 Nov 2026. `Cortex Search services`, `Cortex Analyst services`, `Cortex Agent services`, `object manager group`, `query manager (SQL execution) group`, `semantic manager group` ## Not found in the sources read - A lineage or column-lineage tool in the Snowflake-Labs/mcp README (checked tool and capability list). [Source](https://raw.githubusercontent.com/Snowflake-Labs/mcp/main/README.md). ## The evidence on Snowflake Labs MCP Server, by topic ### Publisher and releases - Snowflake Labs marks its Snowflake-Labs/mcp repository as deprecated and unmaintained, pointing users to the official Snowflake MCP server. (Snowflake Labs's own claim; [source](https://raw.githubusercontent.com/Snowflake-Labs/mcp/main/README.md), read 10 Nov 2026.) - The latest PyPI/GitHub release of snowflake-labs-mcp is v1.4.2, published 15 May 2026 and titled as the release that starts deprecation. (observed in the source; [source](https://github.com/Snowflake-Labs/mcp/releases), read 10 Nov 2026.) - PyPI lists snowflake-labs-mcp 1.4.2 uploaded 15 May 2026 by Snowflake Labs authors. (observed in the source; [source](https://pypi.org/pypi/snowflake-labs-mcp/json), read 10 Nov 2026.) ### Clients, auth and scoping - The server served Cortex Search, Cortex Analyst and Cortex Agent services to MCP clients as tools, configured in a YAML file. (Snowflake Labs's own claim; [source](https://raw.githubusercontent.com/Snowflake-Labs/mcp/main/README.md), read 10 Nov 2026.) - The server honors the RBAC grants of the role used for the connection, so access is scoped to that Snowflake role. (Snowflake Labs's own claim; [source](https://raw.githubusercontent.com/Snowflake-Labs/mcp/main/README.md), read 10 Nov 2026.) ### Access and safety - SQL statement types are allowed or denied per type in a sql_statement_permissions config block. (Snowflake Labs's own claim; [source](https://raw.githubusercontent.com/Snowflake-Labs/mcp/main/README.md), read 10 Nov 2026.) - Authentication supports password, key pair, OAuth, SSO and MFA via the Snowflake connector. (Snowflake Labs's own claim; [source](https://raw.githubusercontent.com/Snowflake-Labs/mcp/main/README.md), read 10 Nov 2026.) ### What it reaches - Covers Snowflake only: Cortex Search, Analyst, Agent, object management and SQL execution. (Snowflake Labs's own claim; [source](https://raw.githubusercontent.com/Snowflake-Labs/mcp/main/README.md), read 10 Nov 2026.) ### Lineage and impact - Legacy tooling included semantic view discovery and querying plus Cortex Analyst, but no lineage tool is listed. (Snowflake Labs's own claim; [source](https://raw.githubusercontent.com/Snowflake-Labs/mcp/main/README.md), read 10 Nov 2026.) ### Definitions, ownership and rules - The legacy server could discover Snowflake Semantic Views and their components, which carry business definitions. (Snowflake Labs's own claim; [source](https://raw.githubusercontent.com/Snowflake-Labs/mcp/main/README.md), read 10 Nov 2026.) ### Freshness and upkeep - Queries run live against Snowflake through the Python connector at call time; no collected or cached metadata store is described. (Snowflake Labs's own claim; [source](https://raw.githubusercontent.com/Snowflake-Labs/mcp/main/README.md), read 10 Nov 2026.) ### Profile - Transports are stdio (default), sse (legacy) and streamable-http; runs via uvx or Docker. (Snowflake Labs's own claim; [source](https://raw.githubusercontent.com/Snowflake-Labs/mcp/main/README.md), read 10 Nov 2026.) ## Read next - [MCP servers for Snowflake](/mcp/snowflake/): 3 listed - [mcp-snowflake-server (community)](/servers/isaacwasserman-mcp-snowflake-server/): also for Snowflake - [Snowflake-managed MCP server](/servers/snowflake-managed-mcp/): also for Snowflake - [Snowflake Labs MCP Server and Snowflake-managed MCP server, side by side](/compare/snowflake-labs-mcp-vs-snowflake-managed-mcp/) - [mcp-snowflake-server (community) and Snowflake Labs MCP Server, side by side](/compare/isaacwasserman-mcp-snowflake-server-vs-snowflake-labs-mcp/) - [The Context Weight Index](/context-weight/) ## Sources - raw.githubusercontent.com/Snowflake-Labs/mcp/main/README.md: https://raw.githubusercontent.com/Snowflake-Labs/mcp/main/README.md (read 10 Nov 2026) - github.com/Snowflake-Labs/mcp/releases: https://github.com/Snowflake-Labs/mcp/releases (read 10 Nov 2026) - pypi.org/pypi/snowflake-labs-mcp/json: https://pypi.org/pypi/snowflake-labs-mcp/json (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/servers/snowflake-labs-mcp/ --- # Snowflake-managed MCP server: 6 tools, remote (hosted), from Snowflake SERVER, read 10 Nov 2026 **The short answer.** Snowflake-managed MCP server is an official MCP server for Snowflake, published by Snowflake and read from its own source on 10 Nov 2026. It runs remote (hosted), and 6 tools are named. Version read: Docs page, MCP revision 25 Nov 2025. Last release: none found on 10 Nov 2026. - **Data tool:** [Snowflake](/mcp/snowflake/) - **Publisher:** Snowflake - **Kind:** Official - **Repository or docs:** [docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-mcp](https://docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-mcp) - **Version or commit:** Docs page, MCP revision 25 Nov 2025 - **Last release:** none found on 10 Nov 2026 - **Transport:** Remote (hosted) - **Auth:** Snowflake OAuth (default) or External OAuth; PAT supported; RBAC per MCP server and per tool. - **Access:** Configurable - **Tools listed:** 6 named in the evidence - **Tool definitions read at:** [docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-mcp](https://docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-mcp) - **Context weight:** Not measured: The docs page defines user-configured tool types (CORTEX_AGENT_RUN, CODING_AGENT, CORTEX_SEARCH_SERVICE_QUERY, CORTEX_ANALYST_MESSAGE, SYSTEM_EXECUTE_SQL, GENERIC) used to build tools in a CREATE MCP SERVER specification; it publishes no fixed list of named tools, so facts.json lists tool types not tools. - **Status:** Documented by Snowflake as the Snowflake-managed MCP server (read 10 Nov 2026); the Snowflake-Labs repository points users here. - **Read:** 10 Nov 2026 ## In one sentence Snowflake's hosted endpoint applies RBAC per server and per tool with OAuth, and serves Cortex Analyst over semantic views, Cortex Search and SQL. ## What Snowflake-managed MCP server exposes 6 tools are named in the evidence read on 10 Nov 2026. `CORTEX_AGENT_RUN`, `CODING_AGENT`, `CORTEX_SEARCH_SERVICE_QUERY`, `CORTEX_ANALYST_MESSAGE`, `SYSTEM_EXECUTE_SQL`, `GENERIC` ## Not found in the sources read - A lineage, column-lineage or object-dependency tool type on the Snowflake-managed MCP server page (tool types list read). [Source](https://docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-mcp). ## The evidence on Snowflake-managed MCP server, by topic ### Publisher and releases - Snowflake documents a Snowflake-managed MCP server that lets agents retrieve data from Snowflake accounts without deploying separate infrastructure. (Snowflake's own claim; [source](https://docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-mcp), read 10 Nov 2026.) - The docs state support for MCP protocol revision 25 Nov 2025; no semantic version number or release date is published on the page. (Snowflake's own claim; [source](https://docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-mcp), read 10 Nov 2026.) ### Clients, auth and scoping - MCP clients discover and invoke Cortex Analyst, Cortex Search and Cortex Agents as tools on the managed server. (Snowflake's own claim; [source](https://docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-mcp), read 10 Nov 2026.) - Access to the MCP server does not give access to tools; privileges are granted per tool, so results depend on the calling role. (Snowflake's own claim; [source](https://docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-mcp), read 10 Nov 2026.) ### Access and safety - Authentication is Snowflake OAuth by default, with optional External OAuth, and role-based access control on the server and tools. (Snowflake's own claim; [source](https://docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-mcp), read 10 Nov 2026.) - Docs list distinct privileges: USAGE on the MCP server plus SELECT on the semantic view, USAGE on the Cortex Search service, agent, or UDF. (Snowflake's own claim; [source](https://docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-mcp), read 10 Nov 2026.) - Docs warn that tool annotations are client hints and not security controls; RBAC is the control. (Snowflake's own claim; [source](https://docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-mcp), read 10 Nov 2026.) - The managed server is available in government regions except Azure US Gov Virginia and has not completed formal FedRAMP assessment. (Snowflake's own claim; [source](https://docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-mcp), read 10 Nov 2026.) ### What it reaches - Serves Snowflake data only; custom tools can wrap UDFs and stored procedures. (Snowflake's own claim; [source](https://docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-mcp), read 10 Nov 2026.) ### Lineage and impact - Tool types are Cortex Agent, coding agent, Cortex Search, Cortex Analyst, SQL execution and generic UDF/procedure tools; none is a lineage tool. (Snowflake's own claim; [source](https://docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-mcp), read 10 Nov 2026.) ### Definitions, ownership and rules - Cortex Analyst tools use semantic views, which hold business definitions, and the server accepts semantic views only, not semantic models. (Snowflake's own claim; [source](https://docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-mcp), read 10 Nov 2026.) ### Freshness and upkeep - Tools run against live Snowflake objects; no metadata snapshot or refresh schedule is described. (Snowflake's own claim; [source](https://docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-mcp), read 10 Nov 2026.) ### Profile - The server is a Snowflake object created with CREATE MCP SERVER, reachable at an account-scoped HTTP URL (remote). (Snowflake's own claim; [source](https://docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-mcp), read 10 Nov 2026.) ## Read next - [MCP servers for Snowflake](/mcp/snowflake/): 3 listed - [mcp-snowflake-server (community)](/servers/isaacwasserman-mcp-snowflake-server/): also for Snowflake - [Snowflake Labs MCP Server](/servers/snowflake-labs-mcp/): also for Snowflake - [Snowflake Labs MCP Server and Snowflake-managed MCP server, side by side](/compare/snowflake-labs-mcp-vs-snowflake-managed-mcp/) - [mcp-snowflake-server (community) and Snowflake-managed MCP server, side by side](/compare/isaacwasserman-mcp-snowflake-server-vs-snowflake-managed-mcp/) - [The Context Weight Index](/context-weight/) ## Sources - docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-mcp: https://docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-mcp (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/servers/snowflake-managed-mcp/ --- # SYNQ MCP (Coalesce Quality): 134 tools, remote (hosted), from SYNQ SERVER, read 10 Nov 2026 **The short answer.** SYNQ MCP (Coalesce Quality) is an official MCP server for SYNQ, published by SYNQ and read from its own source on 10 Nov 2026. It runs remote (hosted), and 134 tools are named. Version read: Hosted, no version published. Last release: none found on 10 Nov 2026. - **Data tool:** [SYNQ](/mcp/synq/) - **Publisher:** SYNQ (docs titled Coalesce Quality) - **Kind:** Official - **Repository or docs:** [docs.synq.io/scout/mcp.md](https://docs.synq.io/scout/mcp.md) - **Version or commit:** Hosted, no version published - **Last release:** none found on 10 Nov 2026 - **Transport:** Remote (hosted) - **Auth:** OAuth 2 with PKCE per user with selectable permission tier (read, write, governance, deploy); legacy workspace token (st-...) as Bearer. - **Access:** Configurable - **Tools listed:** 134 named in the evidence - **Tool definitions read at:** [docs.synq.io/scout/mcp.md](https://docs.synq.io/scout/mcp.md) - **Context weight:** 4,550 tokens for 134 tool definitions (134 with a description), measured from the docs page read - **Status:** Exists on 10 Nov 2026; docs.synq.io now titles itself Coalesce Quality Documentation and its MCP page calls the product Coalesce Quality; regional hosted endpoints EU/US/AU. - **Read:** 10 Nov 2026 ## In one sentence SYNQ's hosted observability and lineage server uses per-user OAuth with tiered consent scopes, and tools hide when a scope is not granted. ## What SYNQ MCP (Coalesce Quality) exposes 134 tools are named in the evidence read on 10 Nov 2026. `add_comment`, `apply_service_account_changes`, `batch_entity_annotations`, `batch_entity_details`, `batch_executions`, `batch_urls`, `browse_folders`, `create_incident`, `create_monitors`, `create_sql_tests`, `create_table_stats_monitor`, `decide_recommendation`, `delete_alert`, `delete_check_categorisation_rule`, `delete_data_product`, `delete_deployment_rule`, `delete_domain`, `delete_monitors`, `delete_owner`, `delete_saved_view`, `delete_sql_test_deployment_rule`, `delete_sql_tests`, `deploy_deployment_rule`, `deploy_sql_test_deployment_rule`, `describe_check_categorisation`, `describe_check_types`, `describe_resolver_ql`, `exclude_from_monitoring`, `execute_monitor`, `folder_of`, `get_alert`, `get_code`, `get_commit_diff`, `get_common_upstream_dependencies`, `get_coverage_summary`, `get_data_product`, `get_database_change_details`, `get_database_coordinates_from_entity`, `get_deployment_rule`, `get_domain`, `get_downstream_dependencies`, `get_entity_access`, `get_entity_details`, `get_entity_from_database_coordinates`, `get_entity_impact`, `get_entity_knowledge`, `get_entity_metrics`, `get_entity_overlays`, `get_immediate_downstream_dependencies`, `get_immediate_upstream_dependencies`, `get_incident_details`, `get_issue`, `get_issue_comments`, `get_issue_details`, `get_issue_impact`, `get_join_patterns`, `get_knowledge_pages`, `get_latest_executions`, `get_lineage`, `get_monitor`, `get_monitor_predictions`, `get_owner`, `get_query_activity`, `get_saved_view`, `get_schema`, `get_sql_test_deployment_rule`, `get_upstream_dependencies`, `get_upstream_sources`, `iam`, `list_access_signals`, `list_alerts`, `list_annotations`, `list_categorised_checks`, `list_check_categories`, `list_check_categorisation_rules`, `list_checks`, `list_commits`, `list_data_product_members`, `list_data_products`, `list_database_changes`, `list_deployment_rules`, `list_domain_members`, `list_domains`, `list_entity_coverage`, `list_executions`, `list_folder_members`, `list_history`, `list_integration_health`, `list_knowledge_pages`, `list_monitor_segments`, `list_monitoring_exclusions`, `list_monitors`, `list_open_incidents`, `list_open_issues`, `list_owners`, `list_ownerships`, `list_platform_users`, `list_recommendations`, `list_saved_views`, `list_service_account_suggestions`, `list_service_accounts`, `list_sql_test_deployment_rules`, `list_sql_tests`, `list_table_stats_monitors`, `list_users`, `plan_service_account_changes`, `preview_check_categorisation_rule`, `preview_deployment_rule`, `preview_sql_test_deployment_rule`, `profile_columns`, `resolve_query`, `run_sql_test_again`, `sample_column_values`, `sample_entities`, `save_alert`, `save_check_categorisation_rule`, `save_data_product`, `save_domain`, `save_owner`, `save_saved_view`, `search_entities`, `send_test_alert`, `set_alert_enabled`, `set_check_category`, `set_issue_status`, `set_monitor_segment_labels`, `set_saved_view_pin`, `submit_feature_request`, `summarise_executions`, `unexclude_from_monitoring`, `update_incident`, `update_monitors`, `update_sql_tests`, `update_table_stats_monitor` ## Not found in the sources read - A GitHub repository, package version or release date for the SYNQ MCP server on the docs page. [Source](https://docs.synq.io/scout/mcp.md). ## The evidence on SYNQ MCP (Coalesce Quality), by topic ### Publisher and releases - The server is hosted by SYNQ at regional URLs (EU mcp.synq.io, US, AU) and needs no workspace setup; no version or release date is published. (SYNQ's own claim; [source](https://docs.synq.io/scout/mcp.md), read 10 Nov 2026.) ### Clients, auth and scoping - OAuth 2 with PKCE means actions are recorded as the real user with a via MCP badge. (SYNQ's own claim; [source](https://docs.synq.io/scout/mcp.md), read 10 Nov 2026.) - The set of tools an assistant sees depends on the permissions the user grants at consent, and the consent screen offers only what the user's role allows. (SYNQ's own claim; [source](https://docs.synq.io/scout/mcp.md), read 10 Nov 2026.) - The iam tool reports the caller's workspace, granted scopes and user. (SYNQ's own claim; [source](https://docs.synq.io/scout/mcp.md), read 10 Nov 2026.) ### Access and safety - Four permission tiers (read-only, write, governance, deploy) each grant only the scopes asked for; read-only makes no changes. (SYNQ's own claim; [source](https://docs.synq.io/scout/mcp.md), read 10 Nov 2026.) - Write tools do not appear in the client if the user granted only read-only access. (SYNQ's own claim; [source](https://docs.synq.io/scout/mcp.md), read 10 Nov 2026.) - Admins can switch off OAuth access to the API for third-party clients, including MCP, with the Disable user API access setting; legacy static workspace tokens (st-...) are also accepted. (SYNQ's own claim; [source](https://docs.synq.io/scout/mcp.md), read 10 Nov 2026.) ### What it reaches - The docs name dbt, SQLMesh, Coalesce, Airflow and warehouse query history as sources the tools draw on. (SYNQ's own claim; [source](https://docs.synq.io/scout/mcp.md), read 10 Nov 2026.) ### Lineage and impact - SYNQ (docs now titled Coalesce Quality) serves the same lineage graph as its lineage view, including column-level lineage where the warehouse supports it. (SYNQ's own claim; [source](https://docs.synq.io/scout/mcp.md), read 10 Nov 2026.) - Lineage tools include get_lineage (up/down/both with depth), upstream and downstream dependencies, common upstream, and upstream sources such as dbt sources. (SYNQ's own claim; [source](https://docs.synq.io/scout/mcp.md), read 10 Nov 2026.) - Get_code returns the latest source code defining an entity, including SQL definitions and dbt models. (SYNQ's own claim; [source](https://docs.synq.io/scout/mcp.md), read 10 Nov 2026.) - Impact tools list the entities and systems affected by an entity or an issue. (SYNQ's own claim; [source](https://docs.synq.io/scout/mcp.md), read 10 Nov 2026.) ### Definitions, ownership and rules - Knowledge tools read entity descriptions and knowledge pages pinned to entities, including owners and certified pages. (SYNQ's own claim; [source](https://docs.synq.io/scout/mcp.md), read 10 Nov 2026.) - The Governance permission tier lets the assistant manage data products, domains, owners, alert routing and check categories. (SYNQ's own claim; [source](https://docs.synq.io/scout/mcp.md), read 10 Nov 2026.) - List_annotations and batch_entity_annotations read annotations carried by entities. (SYNQ's own claim; [source](https://docs.synq.io/scout/mcp.md), read 10 Nov 2026.) ### Freshness and upkeep - Tools return execution history and latest executions for monitors, dbt models and Airflow tasks; join patterns use the last 30 days of warehouse query history. (SYNQ's own claim; [source](https://docs.synq.io/scout/mcp.md), read 10 Nov 2026.) ### Profile - SYNQ's documentation is titled Coalesce Quality Documentation; the relationship between the names is not explained on the pages read. (SYNQ's own claim; [source](https://docs.synq.io/llms.txt), read 10 Nov 2026.) ## Read next - [MCP servers for SYNQ](/mcp/synq/): 1 listed - [Monte Carlo MCP Server and SYNQ MCP (Coalesce Quality), side by side](/compare/monte-carlo-mcp-vs-synq-mcp/) - [The Context Weight Index](/context-weight/) ## Sources - docs.synq.io/scout/mcp.md: https://docs.synq.io/scout/mcp.md (read 10 Nov 2026) - docs.synq.io/llms.txt: https://docs.synq.io/llms.txt (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/servers/synq-mcp/ --- # Tableau MCP: 40 tools, remote and local, from Tableau SERVER, read 10 Nov 2026 **The short answer.** Tableau MCP is an official MCP server for Tableau, published by Tableau and read from its own source on 10 Nov 2026. It runs remote and local, and 40 tools are named. Version read: 6.0.4. Last release: 9 Nov 2026. 25 sourced facts follow. - **Data tool:** [Tableau](/mcp/tableau/) - **Publisher:** Tableau (Salesforce) - **Kind:** Official - **Repository or docs:** [github.com/tableau/tableau-mcp](https://github.com/tableau/tableau-mcp) - **Version or commit:** 6.0.4 - **Last release:** 9 Nov 2026 - **Transport:** Remote and local - **Auth:** Hosted: OAuth 2.1 with each user's own Tableau Cloud identity. Self-hosted: PAT (default), direct-trust (Connected Apps), OAuth, passthrough, or unified access tokens. - **Access:** Configurable - **Tools listed:** 40 named in the evidence - **Tool definitions read at:** [raw.githubusercontent.com/tableau/tableau-mcp/main/src/tools/web/getDatasourceMetadata/getDatasourceMetadata.ts](https://raw.githubusercontent.com/tableau/tableau-mcp/main/src/tools/web/getDatasourceMetadata/getDatasourceMetadata.ts) - **Context weight:** Not measured: the source read on 10 Nov 2026 holds the definitions of 2 of the roughly 40 Tableau MCP tools (get-datasource-metadata and query-datasource), and one of those descriptions is imported from a file that was not read, so a measurement would cover a small part of the server. - **Status:** Exists on 10 Nov 2026 as tableau/tableau-mcp (npm @tableau/mcp-server) and as a hosted service at mcp.tableau.com; the schema file holds two of the tool source files (get-datasource-metadata, query-datasource); the full tool list comes from the docs intro page. - **Read:** 10 Nov 2026 ## In one sentence Tableau's server exposes data source metadata with calculation strings and Pulse metric definitions per signed-in user, plus, on Tableau+, a knowledge tool for lineage and impact. ## What Tableau MCP exposes 40 tools are named in the evidence read on 10 Nov 2026. `list-datasources`, `list-workbooks`, `list-projects`, `list-views`, `list-custom-views`, `get-datasource-metadata`, `get-workbook`, `download-workbook`, `get-flow`, `describe-flow`, `get-flow-task`, `list-flows`, `list-flow-runs`, `list-flow-tasks`, `run-flow`, `run-flow-task`, `cancel-flow-run`, `delete-content`, `get-view-data`, `get-view-image`, `get-custom-view-data`, `get-custom-view-image`, `query-datasource`, `list-all-pulse-metric-definitions`, `list-pulse-metric-definitions-from-definition-ids`, `list-pulse-metrics-from-metric-definition-id`, `list-pulse-metrics-from-metric-ids`, `list-pulse-metric-subscriptions`, `generate-pulse-metric-value-insight-bundle`, `generate-pulse-insight-brief`, `search-content`, `list-extract-refresh-tasks`, `update-cloud-extract-refresh-task`, `list-users`, `update-user`, `query-admin-insights`, `query-knowledge-context`, `inspect-knowledge-context`, `manage-knowledge-context`, `scaffold-data-app` ## Not found in the sources read - Any scheduled or event-driven metadata collection in the Tableau MCP docs intro page; the server calls Tableau REST, Metadata and VizQL APIs per request. [Source](https://tableau.github.io/tableau-mcp/docs/intro). ## The evidence on Tableau MCP, by topic ### Publisher and releases - Tableau publishes tableau-mcp as its official MCP server, and the repo README carries the 'Tableau Supported' support-level badge. (Tableau's own claim; [source](https://tableau.github.io/tableau-mcp/docs/intro), read 10 Nov 2026.) - The latest GitHub release is v6.0.4 (listed first on the releases page). (observed in the source; [source](https://github.com/tableau/tableau-mcp/releases), read 10 Nov 2026.) - The npm package @tableau/mcp-server has dist-tag latest 6.0.4, published 9 Nov 2026, under the Apache-2.0 license. (observed in the source; [source](https://registry.npmjs.org/@tableau/mcp-server), read 10 Nov 2026.) - The README badge states the support level as Tableau Supported. (observed in the source; [source](https://raw.githubusercontent.com/tableau/tableau-mcp/main/README.md), read 10 Nov 2026.) ### Clients, auth and scoping - The hosted Tableau MCP at mcp.tableau.com uses OAuth 2.1 so each user signs in as themselves and per-user Tableau permissions are enforced. (Tableau's own claim; [source](https://github.com/tableau/tableau-mcp), read 10 Nov 2026.) - Authentication to the Tableau REST APIs is selectable between pat (default), direct-trust, oauth and passthrough. (Tableau's own claim; [source](https://tableau.github.io/tableau-mcp/docs/configuration/mcp-config/env-vars), read 10 Nov 2026.) - The Tableau docs state the authenticated user must have access to the published data sources the agent will use. (Tableau's own claim; [source](https://tableau.github.io/tableau-mcp/docs/configuration/mcp-config/authentication), read 10 Nov 2026.) ### Access and safety - Tools can be limited with INCLUDE_TOOLS or EXCLUDE_TOOLS environment variables (comma-separated tool or group names). (Tableau's own claim; [source](https://tableau.github.io/tableau-mcp/docs/configuration/mcp-config/env-vars), read 10 Nov 2026.) - Mutating flow tools (run-flow, run-flow-task, cancel-flow-run) are described as preview-and-confirm tools, and delete-content is admin-only and two-phase. (Tableau's own claim; [source](https://tableau.github.io/tableau-mcp/docs/intro), read 10 Nov 2026.) - The query-datasource tool is annotated readOnlyHint true and destructiveHint false in source. (observed in the source; [source](https://raw.githubusercontent.com/tableau/tableau-mcp/main/src/tools/web/queryDatasource/queryDatasource.ts), read 10 Nov 2026.) - Telemetry defaults to the noop provider. (Tableau's own claim; [source](https://tableau.github.io/tableau-mcp/docs/configuration/mcp-config/env-vars), read 10 Nov 2026.) ### What it reaches - The server works with Tableau Cloud and Tableau Server of any edition, with some tools gated by SKU. (Tableau's own claim; [source](https://tableau.github.io/tableau-mcp/docs/intro), read 10 Nov 2026.) - Tools span published data sources, workbooks, views, custom views, projects, Prep flows, Pulse metrics, extract refresh tasks, users and Admin Insights. (Tableau's own claim; [source](https://tableau.github.io/tableau-mcp/docs/intro), read 10 Nov 2026.) - The server speaks stdio by default and HTTP as an alternative transport. (Tableau's own claim; [source](https://tableau.github.io/tableau-mcp/docs/configuration/mcp-config/env-vars), read 10 Nov 2026.) ### Lineage and impact - Tableau's get-datasource-metadata tool returns table relationships, field descriptions, roles and types, calculation strings and parameters for a data source (table, column and calculation-definition level, not transformation lineage). (Tableau's own claim; [source](https://tableau.github.io/tableau-mcp/docs/intro), read 10 Nov 2026.) - The describe-flow tool summarizes a Tableau Prep flow's design including inputs, outputs, steps, lineage and connections, on all SKUs. (Tableau's own claim; [source](https://tableau.github.io/tableau-mcp/docs/intro), read 10 Nov 2026.) - On Tableau+ only, query-knowledge-context queries governed definitions, relationships, lineage, impact and graph sources through one read-only tool. (Tableau's own claim; [source](https://tableau.github.io/tableau-mcp/docs/intro), read 10 Nov 2026.) - The server's get-datasource-metadata tool enriches VizQL Data Service results with Tableau's Metadata API and only returns the richer descriptions for published data sources. (observed in the source; [source](https://raw.githubusercontent.com/tableau/tableau-mcp/main/src/tools/web/getDatasourceMetadata/getDatasourceMetadata.ts), read 10 Nov 2026.) ### Definitions, ownership and rules - Tableau MCP serves Pulse metric definitions and the Pulse insights engine to agents so metrics are answered from governed definitions. (Tableau's own claim; [source](https://tableau.github.io/tableau-mcp/docs/intro), read 10 Nov 2026.) - On Tableau+ only, manage-knowledge-context creates, updates or deletes customer-governed semantic context, and inspect-knowledge-context reports graph status and improvement suggestions. (Tableau's own claim; [source](https://tableau.github.io/tableau-mcp/docs/intro), read 10 Nov 2026.) - The intro page positions the server on reusing curated data models and metadata so agent answers are grounded in pre-built data semantics. (Tableau's own claim; [source](https://tableau.github.io/tableau-mcp/docs/intro), read 10 Nov 2026.) ### Freshness and upkeep - Query-datasource runs VizQL Data Service queries against published data sources at call time, so answers reflect live data rather than a collected copy. (Tableau's own claim; [source](https://tableau.github.io/tableau-mcp/docs/intro), read 10 Nov 2026.) ### Profile - The docs site footer credits Salesforce, Inc. as the copyright holder of Tableau MCP. (Tableau's own claim; [source](https://tableau.github.io/tableau-mcp/docs/intro), read 10 Nov 2026.) - The package can run via npx (Node 22.7.5 or later), Docker, Heroku, or as Node single-executable binaries; a separate Tableau Desktop authoring server is not in the npm package. (Tableau's own claim; [source](https://github.com/tableau/tableau-mcp), read 10 Nov 2026.) ## Read next - [MCP servers for Tableau](/mcp/tableau/): 1 listed - [Power BI Modeling MCP Server and Tableau MCP, side by side](/compare/powerbi-modeling-mcp-vs-tableau-mcp/) - [Looker tools in MCP Toolbox for Databases and Tableau MCP, side by side](/compare/looker-mcp-toolbox-vs-tableau-mcp/) - [The Context Weight Index](/context-weight/) ## Sources - tableau.github.io/tableau-mcp/docs/intro: https://tableau.github.io/tableau-mcp/docs/intro (read 10 Nov 2026) - github.com/tableau/tableau-mcp/releases: https://github.com/tableau/tableau-mcp/releases (read 10 Nov 2026) - registry.npmjs.org/@tableau/mcp-server: https://registry.npmjs.org/@tableau/mcp-server (read 10 Nov 2026) - raw.githubusercontent.com/tableau/tableau-mcp/main/README.md: https://raw.githubusercontent.com/tableau/tableau-mcp/main/README.md (read 10 Nov 2026) - raw.githubusercontent.com/tableau/tableau-mcp/main/src/tools/web/getDatasourceMetadata/getDatasourceMetadata.ts: https://raw.githubusercontent.com/tableau/tableau-mcp/main/src/tools/web/getDatasourceMetadata/getDatasourceMetadata.ts (read 10 Nov 2026) - github.com/tableau/tableau-mcp: https://github.com/tableau/tableau-mcp (read 10 Nov 2026) - tableau.github.io/tableau-mcp/docs/configuration/mcp-config/env-vars: https://tableau.github.io/tableau-mcp/docs/configuration/mcp-config/env-vars (read 10 Nov 2026) - tableau.github.io/tableau-mcp/docs/configuration/mcp-config/authentication: https://tableau.github.io/tableau-mcp/docs/configuration/mcp-config/authentication (read 10 Nov 2026) - raw.githubusercontent.com/tableau/tableau-mcp/main/src/tools/web/queryDatasource/queryDatasource.ts: https://raw.githubusercontent.com/tableau/tableau-mcp/main/src/tools/web/queryDatasource/queryDatasource.ts (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/servers/tableau-mcp/ --- # ThoughtSpot MCP Server: 20 tools, remote (hosted), from ThoughtSpot SERVER, read 10 Nov 2026 **The short answer.** ThoughtSpot MCP Server is an official MCP server for ThoughtSpot, published by ThoughtSpot and read from its own source on 10 Nov 2026. It runs remote (hosted), and 20 tools are named. Version read: 0.5.0 (npm and package.json); main at commit 820bf16f6aa3. Last release: 6 Jun 2025. - **Data tool:** [ThoughtSpot](/mcp/thoughtspot/) - **Publisher:** ThoughtSpot - **Kind:** Official - **Repository or docs:** [github.com/thoughtspot/mcp-server](https://github.com/thoughtspot/mcp-server) - **Version or commit:** 0.5.0 (npm and package.json); main at commit 820bf16f6aa3 - **Last release:** 6 Jun 2025 - **Transport:** Remote (hosted) - **Auth:** OAuth to a ThoughtSpot instance (dynamic client registration, or manually registered client id and secret); bearer-token and /sse paths also exist; the user chooses or is bound to a ThoughtSpot instance URL. - **Access:** Read and write - **Tools listed:** 20 named in the evidence - **Tool definitions read at:** [raw.githubusercontent.com/thoughtspot/mcp-server/main/src/servers/tool-definitions.ts](https://raw.githubusercontent.com/thoughtspot/mcp-server/main/src/servers/tool-definitions.ts) - **Context weight:** at least 2,149 tokens for 20 tool definitions (20 with a description), measured from the source code read - **Status:** Exists on 10 Nov 2026 as thoughtspot/mcp-server, run as a Cloudflare Worker (hosted at agent.thoughtspot.app) with a local stdio entry in package.json. The GitHub repo has no releases or tags; the npm package last published 6 Jun 2025 although main has moved on, so last_release comes from npm. - **Read:** 10 Nov 2026 ## In one sentence ThoughtSpot's server returns Spotter answers, searches objects with owner, tag and verified filters, and creates dashboards and models within the user's privileges. ## What ThoughtSpot MCP Server exposes 20 tools are named in the evidence read on 10 Nov 2026. `ping`, `getRelevantQuestions`, `getAnswer`, `createLiveboard`, `getDataSourceSuggestions`, `check_connectivity`, `search_objects`, `get_data`, `create_analysis_session`, `send_session_message`, `get_session_updates`, `create_dashboard`, `modify_dashboard`, `get_dashboard_status`, `list_orgs`, `switch_org`, `create_model_session`, `send_model_message`, `get_model_updates`, `finalize_model` ## Not found in the sources read - A lineage tool in tool-definitions.ts (no column, upstream-table or transformation lineage tool among the 20 tools). [Source](https://raw.githubusercontent.com/thoughtspot/mcp-server/main/src/servers/tool-definitions.ts). - Scheduled or event-driven metadata collection in the ThoughtSpot MCP README; tools query the instance per request. [Source](https://raw.githubusercontent.com/thoughtspot/mcp-server/main/README.md). ## The evidence on ThoughtSpot MCP Server, by topic ### Publisher and releases - ThoughtSpot publishes the Spotter MCP Server in its own GitHub organization. (ThoughtSpot's own claim; [source](https://github.com/thoughtspot/mcp-server), read 10 Nov 2026.) - The GitHub releases page for thoughtspot/mcp-server shows no releases. (observed in the source; [source](https://github.com/thoughtspot/mcp-server/releases), read 10 Nov 2026.) - The npm package @thoughtspot/mcp-server has latest 0.5.0, published 6 Jun 2025, under a ThoughtSpot end user license agreement; repo package.json also reads 0.5.0. (observed in the source; [source](https://registry.npmjs.org/@thoughtspot/mcp-server), read 10 Nov 2026.) ### Clients, auth and scoping - Spotter 'handles authentication securely, and returns live answers from your own data' per the README. (ThoughtSpot's own claim; [source](https://raw.githubusercontent.com/thoughtspot/mcp-server/main/README.md), read 10 Nov 2026.) - The model tools are offered as a group and gated on the user's data-modeling privileges, and get_data is gated on the data-download privilege at runtime. (observed in the source; [source](https://raw.githubusercontent.com/thoughtspot/mcp-server/main/src/servers/tool-definitions.ts), read 10 Nov 2026.) - OAuth client registration can be tied to a ThoughtSpot instance so users do not enter the instance URL. (ThoughtSpot's own claim; [source](https://raw.githubusercontent.com/thoughtspot/mcp-server/main/README.md), read 10 Nov 2026.) ### Access and safety - Search_objects and get_data are annotated readOnlyHint true and destructiveHint false. (observed in the source; [source](https://raw.githubusercontent.com/thoughtspot/mcp-server/main/src/servers/tool-definitions.ts), read 10 Nov 2026.) - Write tools exist: createLiveboard, create_dashboard, modify_dashboard and the model tools create or change content. (observed in the source; [source](https://raw.githubusercontent.com/thoughtspot/mcp-server/main/src/servers/tool-definitions.ts), read 10 Nov 2026.) - INSTANCE_ALLOWLIST_BY_HOST restricts which ThoughtSpot instances each Worker domain accepts, returning 403 for others. (ThoughtSpot's own claim; [source](https://raw.githubusercontent.com/thoughtspot/mcp-server/main/README.md), read 10 Nov 2026.) ### What it reaches - The tool set includes connectivity and auth checks (ping, check_connectivity) and org handling tools list_orgs and switch_org, so one server can reach multiple ThoughtSpot Orgs. (observed in the source; [source](https://raw.githubusercontent.com/thoughtspot/mcp-server/main/src/servers/tool-definitions.ts), read 10 Nov 2026.) - The README lists /mcp, /sse, /token/* and /bearer/* endpoints on the Worker. (ThoughtSpot's own claim; [source](https://raw.githubusercontent.com/thoughtspot/mcp-server/main/README.md), read 10 Nov 2026.) ### Lineage and impact - Search_objects finds Answers, Liveboards and Worksheets with filters for types, owner, tag, modified_since and verified_only, returning object metadata (title, description, query tokens, tags). (observed in the source; [source](https://raw.githubusercontent.com/thoughtspot/mcp-server/main/src/servers/tool-definitions.ts), read 10 Nov 2026.) - The tool source warns that description and query tokens are authoring metadata that can be stale, and advises grounding explanations in the object's data via get_data. (observed in the source; [source](https://raw.githubusercontent.com/thoughtspot/mcp-server/main/src/servers/tool-definitions.ts), read 10 Nov 2026.) ### Definitions, ownership and rules - Search_objects results include a Verified flag and Owner, and tags, so certification and ownership signals reach the agent. (observed in the source; [source](https://raw.githubusercontent.com/thoughtspot/mcp-server/main/src/servers/tool-definitions.ts), read 10 Nov 2026.) - A V3 'Spotter Model' tool group (create_model_session, send_model_message, get_model_updates, finalize_model) lets an agent create and edit data models. (observed in the source; [source](https://raw.githubusercontent.com/thoughtspot/mcp-server/main/src/servers/tool-definitions.ts), read 10 Nov 2026.) ### Freshness and upkeep - Get_data re-runs a saved Answer or Liveboard against current data, so rows reflect the latest data. (observed in the source; [source](https://raw.githubusercontent.com/thoughtspot/mcp-server/main/src/servers/tool-definitions.ts), read 10 Nov 2026.) ### Profile - The README links a free ThoughtSpot trial and a hosted client registration page at agent.thoughtspot.app. (ThoughtSpot's own claim; [source](https://raw.githubusercontent.com/thoughtspot/mcp-server/main/README.md), read 10 Nov 2026.) ## Read next - [MCP servers for ThoughtSpot](/mcp/thoughtspot/): 1 listed - [Sigma MCP server and ThoughtSpot MCP Server, side by side](/compare/sigma-mcp-vs-thoughtspot-mcp-server/) - [The Context Weight Index](/context-weight/) ## Sources - github.com/thoughtspot/mcp-server: https://github.com/thoughtspot/mcp-server (read 10 Nov 2026) - github.com/thoughtspot/mcp-server/releases: https://github.com/thoughtspot/mcp-server/releases (read 10 Nov 2026) - registry.npmjs.org/@thoughtspot/mcp-server: https://registry.npmjs.org/@thoughtspot/mcp-server (read 10 Nov 2026) - raw.githubusercontent.com/thoughtspot/mcp-server/main/src/servers/tool-definitions.ts: https://raw.githubusercontent.com/thoughtspot/mcp-server/main/src/servers/tool-definitions.ts (read 10 Nov 2026) - raw.githubusercontent.com/thoughtspot/mcp-server/main/README.md: https://raw.githubusercontent.com/thoughtspot/mcp-server/main/README.md (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/servers/thoughtspot-mcp-server/ --- # One MCP server for Airbyte: Airbyte MCP (hosted), from Airbyte MCP SERVERS FOR A DATA TOOL, read 10 Nov 2026 **The short answer.** One MCP server for Airbyte is listed: Airbyte MCP (hosted), official, published by Airbyte and read on 10 Nov 2026. It runs remote (hosted) and names 50 tools; the version read is Hosted service (no versioned release); PyAirbyte package 0.71.3 ships an experimental local airbyte.mcp server. - **Data tool:** [Airbyte](/mcp/airbyte/) - **Publisher:** Airbyte - **Kind:** Official - **Repository or docs:** [docs.airbyte.com/platform/airbyte-mcp](https://docs.airbyte.com/platform/airbyte-mcp) - **Version or commit:** Hosted service (no versioned release); PyAirbyte package 0.71.3 ships an experimental local airbyte.mcp server - **Last release:** none found on 10 Nov 2026 - **Transport:** Remote (hosted) - **Auth:** OAuth (Streamable HTTP) as the signed-in Airbyte user, or an Airbyte application for non-browser clients; tools run with that user's Airbyte role. - **Access:** Configurable - **Tools listed:** 50 named in the evidence - **Tool definitions read at:** [docs.airbyte.com/platform/airbyte-mcp/tools.md](https://docs.airbyte.com/platform/airbyte-mcp/tools.md) - **Context weight:** 870 tokens for 50 tool definitions (50 with a description), measured from the docs page read - **Status:** Exists on 10 Nov 2026 at mcp.airbyte.com/mcp for Airbyte Cloud and Flex; self-managed Core can run it locally with known issues. A separate PyAirbyte (airbyte.mcp) server and an airbyte-agent-connectors SDK also exist; this file covers the documented Airbyte MCP. - **Read:** 10 Nov 2026 ## Airbyte MCP (hosted) Airbyte's hosted server lets an agent build, run and troubleshoot pipelines as the signed-in user, with a documented read-only mode; it serves pipeline and catalog state. `get_default_workspace_context`, `list_organizations`, `describe_organization`, `get_organization_billing_status`, `list_workspaces`, `describe_workspace`, `set_default_workspace`, `list_connectors`, `describe_connector`, `deploy_connector`, `deploy_noop_destination`, `check_connector`, `rename_connector`, `update_connector_config`, `permanently_delete_connector`, `list_connections`, `describe_connection`, `create_connection`, `update_connection`, `rename_connection`, `set_connection_selected_streams`, `set_connection_table_prefix`, `discover_source_catalog`, `get_connection_artifact`, `permanently_delete_connection`, `run_connection_sync`, `cancel_connection_sync`, `get_connection_sync_status`, `list_connection_sync_jobs`, `get_connection_sync_logs`, `show_workspace_sync_status`, `show_connection_sync_history`, `show_connector_definitions`, `list_custom_source_definitions`, `get_custom_source_definition`, `get_connector_builder_draft_manifest`, `publish_custom_source_definition`, `update_custom_source_definition`, `permanently_delete_custom_source_definition`, `get_external_access_skill_docs`, `execute_external_api_query`, `execute_external_sql_query`, `execute_external_search_query`, `get_search_index_status`, `request_search_index_backfill`, `list_connector_definitions`, `get_connector_definition`, `get_connector_definition_version_history`, `get_api_docs_urls`, `search_airbyte_knowledge_sources` - None found on 10 Nov 2026: Any lineage tool in the Airbyte MCP tools page (column, transformation or BI lineage). - None found on 10 Nov 2026: Glossary, ownership or classification tools in the Airbyte MCP tools page. - None found on 10 Nov 2026: Scheduled or event-driven metadata collection described in the Airbyte MCP docs; tools query on demand. ## Read next - [Airbyte MCP (hosted), the full entry](/servers/airbyte-mcp/): 21 facts - [Airbyte MCP (hosted) and Fivetran MCP Server, side by side](/compare/airbyte-mcp-vs-fivetran-mcp/) - [Ingestion connectors](/categories/ingestion/) ## Sources - docs.airbyte.com/platform/airbyte-mcp.md: https://docs.airbyte.com/platform/airbyte-mcp.md (read 10 Nov 2026) - docs.airbyte.com/platform/airbyte-mcp/install.md: https://docs.airbyte.com/platform/airbyte-mcp/install.md (read 10 Nov 2026) - pypi.org/pypi/airbyte/json: https://pypi.org/pypi/airbyte/json (read 10 Nov 2026) - docs.airbyte.com/llms.txt: https://docs.airbyte.com/llms.txt (read 10 Nov 2026) - docs.airbyte.com/platform/airbyte-mcp/tools.md: https://docs.airbyte.com/platform/airbyte-mcp/tools.md (read 10 Nov 2026) - docs.airbyte.com/platform/airbyte-mcp/advanced-configuration.md: https://docs.airbyte.com/platform/airbyte-mcp/advanced-configuration.md (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/mcp/airbyte/ --- # One MCP server for Alation: Alation AI Agent SDK MCP server, from Alation MCP SERVERS FOR A DATA TOOL, read 10 Nov 2026 **The short answer.** One MCP server for Alation is listed: Alation AI Agent SDK MCP server, official, published by Alation and read on 10 Nov 2026. It runs remote and local and names 12 tools; the version read is 1.0.0. Last release: 26 Oct 2026. Auth: OAuth client credentials of a service account (role-limited) or user refresh token. - **Data tool:** [Alation](/mcp/alation/) - **Publisher:** Alation - **Kind:** Official - **Repository or docs:** [github.com/Alation/alation-ai-agent-sdk](https://github.com/Alation/alation-ai-agent-sdk) - **Version or commit:** 1.0.0 - **Last release:** 26 Oct 2026 - **Transport:** Remote and local - **Auth:** OAuth client credentials of a service account (role-limited) or user refresh token; remote MCP at /ai/mcp needs client_id and client_secret (no dynamic client registration). - **Access:** Not stated in the source read - **Tools listed:** 12 named in the evidence - **Tool definitions read at:** [github.com/Alation/alation-ai-agent-sdk/tree/main/guides](https://github.com/Alation/alation-ai-agent-sdk/tree/main/guides) - **Context weight:** 69 tokens for 4 tool definitions (3 with a description), measured from the docs page read - **Status:** Exists on 10 Nov 2026; alation-ai-agent-mcp 1.0.0 on PyPI and GitHub release v1.0.0 dated 26 Oct 2026. Access is read-only by documentation of removed write tools, not by a published read-only switch. - **Read:** 10 Nov 2026 ## Alation AI Agent SDK MCP server Alation's server offers a lineage tool and a catalog search agent, and its remote form hides tools the current user cannot use. `alation_context`, `bulk_retrieval`, `data_quality_tool`, `get_custom_fields_definitions`, `get_data_products`, `get_data_sources_tool`, `get_signature_creation_instructions`, `lineage`, `catalog_context_search_agent`, `query_flow_agent`, `sql_query_agent`, `custom_agent` - None found on 10 Nov 2026: Documentation that the lineage tool returns transformation SQL/logic or BI-field/metric nodes; the tool page lists graph nodes filtered by object type only. - None found on 10 Nov 2026: A list of warehouses, transformation or BI tools in the Alation agent SDK README (coverage follows the Alation catalog). - None found on 10 Nov 2026: A freshness or collection-cadence statement in the Alation agent SDK README or lineage guide. ## Read next - [Alation AI Agent SDK MCP server, the full entry](/servers/alation-mcp/): 19 facts - [Alation AI Agent SDK MCP server and Collibra MCP Server (chip), side by side](/compare/alation-mcp-vs-collibra-mcp/) - [Catalogs and metadata](/categories/catalogs/) ## Sources - github.com/Alation/alation-ai-agent-sdk/blob/main/README.md: https://github.com/Alation/alation-ai-agent-sdk/blob/main/README.md (read 10 Nov 2026) - github.com/Alation/alation-ai-agent-sdk/blob/main/guides/authentication.md: https://github.com/Alation/alation-ai-agent-sdk/blob/main/guides/authentication.md (read 10 Nov 2026) - pypi.org/pypi/alation-ai-agent-mcp/json: https://pypi.org/pypi/alation-ai-agent-mcp/json (read 10 Nov 2026) - github.com/Alation/alation-ai-agent-sdk/releases: https://github.com/Alation/alation-ai-agent-sdk/releases (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/mcp/alation/ --- # One MCP server for Amazon Redshift: Amazon Redshift MCP Server (AWS Labs), from AWS Labs MCP SERVERS FOR A DATA TOOL, read 10 Nov 2026 **The short answer.** One MCP server for Amazon Redshift is listed: Amazon Redshift MCP Server (AWS Labs), official, labs org, published by AWS Labs and read on 10 Nov 2026. It runs local (stdio) and names 8 tools; the version read is 0.1.3. Last release: 9 Nov 2026. - **Data tool:** [Amazon Redshift](/mcp/redshift/) - **Publisher:** AWS Labs - **Kind:** Official, Labs org - **Repository or docs:** [github.com/awslabs/mcp/tree/main/src/redshift-mcp-server](https://github.com/awslabs/mcp/tree/main/src/redshift-mcp-server) - **Version or commit:** 0.1.3 - **Last release:** 9 Nov 2026 - **Transport:** Local (stdio) - **Auth:** AWS credentials (profile/env) with IAM and Redshift Data API; database user privileges. - **Access:** Configurable - **Tools listed:** 8 named in the evidence - **Tool definitions read at:** [raw.githubusercontent.com/awslabs/mcp/main/src/redshift-mcp-server/README.md](https://raw.githubusercontent.com/awslabs/mcp/main/src/redshift-mcp-server/README.md) - **Context weight:** 238 tokens for 8 tool definitions (8 with a description), measured from the README read - **Status:** Exists in awslabs/mcp; PyPI 0.1.3 uploaded 9 Nov 2026 (read 10 Nov 2026). - **Read:** 10 Nov 2026 ## Amazon Redshift MCP Server (AWS Labs) The AWS Labs server is read-only by default with an explicit read-write opt-in, confirmation prompts and metadata discovery over the Redshift Data API. `list_clusters`, `list_databases`, `list_schemas`, `list_tables`, `list_columns`, `execute_query`, `review_cluster`, `retrieve_skill_file` - None found on 10 Nov 2026: A lineage tool in the Redshift MCP server tool list (eight tools). ## Read next - [Amazon Redshift MCP Server (AWS Labs), the full entry](/servers/awslabs-redshift-mcp-server/): 13 facts - [Warehouses](/categories/warehouses/) ## Sources - pypi.org/pypi/awslabs.redshift-mcp-server/json: https://pypi.org/pypi/awslabs.redshift-mcp-server/json (read 10 Nov 2026) - raw.githubusercontent.com/awslabs/mcp/main/src/redshift-mcp-server/README.md: https://raw.githubusercontent.com/awslabs/mcp/main/src/redshift-mcp-server/README.md (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/mcp/redshift/ --- # One MCP server for Atlan: Atlan MCP Server, from Atlan MCP SERVERS FOR A DATA TOOL, read 10 Nov 2026 **The short answer.** One MCP server for Atlan is listed: Atlan MCP Server, official, published by Atlan and read on 10 Nov 2026. It runs remote (hosted) and names 41 tools; the version read is Hosted at mcp.atlan.com/mcp (unversioned); local atlan-mcp-server 0.3.3 (deprecated). Last release: 17 Feb 2026. - **Data tool:** [Atlan](/mcp/atlan/) - **Publisher:** Atlan - **Kind:** Official - **Repository or docs:** [docs.atlan.com/product/capabilities/atlan-ai/references/mcp-tools](https://docs.atlan.com/product/capabilities/atlan-ai/references/mcp-tools) - **Version or commit:** Hosted at mcp.atlan.com/mcp (unversioned); local atlan-mcp-server 0.3.3 (deprecated) - **Last release:** 17 Feb 2026 - **Transport:** Remote (hosted) - **Auth:** OAuth per user (calls run as the signed-in user) or Atlan API key as Bearer token (shared service identity); local server uses API key + base URL. - **Access:** Configurable - **Tools listed:** 41 named in the evidence - **Tool definitions read at:** [docs.atlan.com/product/capabilities/atlan-ai/references/mcp-tools](https://docs.atlan.com/product/capabilities/atlan-ai/references/mcp-tools) - **Context weight:** 1,770 tokens for 42 tool definitions (42 with a description), measured from the docs page read - **Status:** Exists on 10 Nov 2026: hosted Atlan MCP (41 tools, page updated 27 Oct 2026); the local atlan-mcp-server in atlanhq/agent-toolkit is marked deprecated in its README. - **Read:** 10 Nov 2026 ## Atlan MCP Server Atlan's hosted per-tenant server runs as the signed-in user, scopes context by persona, offers an optional read-only mode and proposes writes for approval. `semantic_search`, `ask`, `fluent_search`, `fluent_search_schema`, `search_assets`, `get_assets`, `search_atlan_docs`, `resolve_metadata`, `get_groups`, `get_users`, `describe_asset_type`, `traverse_lineage`, `query_assets`, `update_assets`, `update_custom_metadata`, `add_atlan_tags`, `remove_atlan_tag`, `manage_announcements`, `create_custom_metadata_set`, `add_attributes_to_cm_set`, `remove_attributes_from_cm_set`, `remove_custom_metadata`, `delete_custom_metadata_set`, `manage_asset_lifecycle`, `create_glossaries`, `create_glossary_categories`, `create_glossary_terms`, `create_domains`, `create_data_products`, `create_dq_rules`, `update_dq_rules`, `schedule_dq_rules`, `delete_dq_rules`, `read_artifact`, `search_knowledge`, `list_knowledge`, `read_knowledge_file`, `create_knowledge_folder`, `save_knowledge_content`, `get_knowledge_upload_url`, `register_knowledge_file` - None found on 10 Nov 2026: A statement on the MCP tools page that traverse_lineage returns column-level edges or transformation SQL. ## Read next - [Atlan MCP Server, the full entry](/servers/atlan-mcp/): 22 facts - [Atlan MCP Server and DataHub MCP Server, side by side](/compare/atlan-mcp-vs-datahub-mcp/) - [Atlan MCP Server and Euno MCP, side by side](/compare/atlan-mcp-vs-euno-mcp/) - [Catalogs and metadata](/categories/catalogs/) ## Sources - docs.atlan.com/product/capabilities/atlan-ai/how-tos/remote-mcp-overview: https://docs.atlan.com/product/capabilities/atlan-ai/how-tos/remote-mcp-overview (read 10 Nov 2026) - docs.atlan.com/product/capabilities/atlan-ai/references/mcp-security: https://docs.atlan.com/product/capabilities/atlan-ai/references/mcp-security (read 10 Nov 2026) - docs.atlan.com/product/capabilities/atlan-ai/references/mcp-tools: https://docs.atlan.com/product/capabilities/atlan-ai/references/mcp-tools (read 10 Nov 2026) - github.com/atlanhq/agent-toolkit/blob/main/modelcontextprotocol/README.md: https://github.com/atlanhq/agent-toolkit/blob/main/modelcontextprotocol/README.md (read 10 Nov 2026) - github.com/atlanhq/agent-toolkit/blob/main/CHANGELOG.md: https://github.com/atlanhq/agent-toolkit/blob/main/CHANGELOG.md (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/mcp/atlan/ --- # 2 MCP servers for BigQuery, read from their own sources MCP SERVERS FOR A DATA TOOL, read 10 Nov 2026 **The short answer.** 2 MCP servers for BigQuery are listed, each read from its own source on 10 Nov 2026: BigQuery remote MCP server (official, remote (hosted)); MCP Toolbox for Databases (BigQuery tools) (official, remote and local). Between them they name 18 tools. The table puts the same fields side by side. | Field | BigQuery remote MCP server | MCP Toolbox for Databases (BigQuery tools) | | --- | --- | --- | | Data tool | [BigQuery](/mcp/bigquery/) | [BigQuery](/mcp/bigquery/) | | Publisher | Google Cloud | Google (googleapis) | | Kind | Official | Official | | Repository or docs | [docs.cloud.google.com/bigquery/docs/use-bigquery-mcp](https://docs.cloud.google.com/bigquery/docs/use-bigquery-mcp) | [github.com/googleapis/mcp-toolbox](https://github.com/googleapis/mcp-toolbox) | | Version or commit | Docs reference last updated 3 Nov 2026 | V1.16.0 | | Last release | none found on 10 Nov 2026 | 5 Nov 2026 | | Transport | Remote (hosted) | Remote and local | | Auth | OAuth 2.0 with IAM (Google Cloud identities, OAuth client, or agent identity). | Application Default Credentials, client OAuth token forwarding, or service account impersonation. | | Access | Configurable | Configurable | | Tools listed | 9 named in the evidence | 9 named in the evidence | | Tool definitions read at | [docs.cloud.google.com/bigquery/docs/reference/mcp](https://docs.cloud.google.com/bigquery/docs/reference/mcp) | [raw.githubusercontent.com/googleapis/mcp-toolbox/main/docs/en/integrations/bigquery/prebuilt-configs/bigquery.md](https://raw.githubusercontent.com/googleapis/mcp-toolbox/main/docs/en/integrations/bigquery/prebuilt-configs/bigquery.md) | | Context weight | 1,487 tokens for 9 tool definitions (9 with a description), measured from the docs page read | 156 tokens for 9 tool definitions (9 with a description), measured from the docs page read | | Status | Exists; Google-managed endpoint, reference page last updated 3 Nov 2026 (read 10 Nov 2026). | Repo googleapis/genai-toolbox now lives at googleapis/mcp-toolbox (read 10 Nov 2026); BigQuery tools documented there. | | Read | 10 Nov 2026 | 10 Nov 2026 | 2 servers for BigQuery, the same fields in the same order, read 10 Nov 2026. ## BigQuery remote MCP server Google Cloud's managed endpoint gives IAM-scoped access to a SELECT-only query tool and dataset and table metadata tools. `list_dataset_ids`, `get_dataset_info`, `list_table_ids`, `get_table_info`, `execute_sql_readonly`, `execute_sql`, `get_query_results`, `cancel_job`, `get_job` - None found on 10 Nov 2026: A lineage or Dataplex lineage tool in the BigQuery MCP tools reference (nine tools listed). ## MCP Toolbox for Databases (BigQuery tools) Google's open-source Toolbox offers BigQuery SQL and metadata tools with a read-only mode, dataset allowlists and Knowledge Catalog search. `analyze_contribution`, `ask_data_insights`, `execute_sql`, `forecast`, `get_dataset_info`, `get_table_info`, `list_dataset_ids`, `list_table_ids`, `search_catalog` - None found on 10 Nov 2026: A lineage tool in the BigQuery prebuilt tool list. ## Read next - [BigQuery remote MCP server, the full entry](/servers/bigquery-remote-mcp/): 13 facts - [MCP Toolbox for Databases (BigQuery tools), the full entry](/servers/mcp-toolbox-bigquery/): 13 facts - [BigQuery remote MCP server and MCP Toolbox for Databases (BigQuery tools), side by side](/compare/bigquery-remote-mcp-vs-mcp-toolbox-bigquery/) - [Warehouses](/categories/warehouses/) ## Sources - docs.cloud.google.com/bigquery/docs/use-bigquery-mcp: https://docs.cloud.google.com/bigquery/docs/use-bigquery-mcp (read 10 Nov 2026) - docs.cloud.google.com/bigquery/docs/reference/mcp: https://docs.cloud.google.com/bigquery/docs/reference/mcp (read 10 Nov 2026) - raw.githubusercontent.com/googleapis/mcp-toolbox/main/README.md: https://raw.githubusercontent.com/googleapis/mcp-toolbox/main/README.md (read 10 Nov 2026) - github.com/googleapis/mcp-toolbox/releases: https://github.com/googleapis/mcp-toolbox/releases (read 10 Nov 2026) - raw.githubusercontent.com/googleapis/mcp-toolbox/main/docs/en/integrations/bigquery/prebuilt-configs/bigquery.md: https://raw.githubusercontent.com/googleapis/mcp-toolbox/main/docs/en/integrations/bigquery/prebuilt-configs/bigquery.md (read 10 Nov 2026) - raw.githubusercontent.com/googleapis/mcp-toolbox/main/docs/en/integrations/bigquery/source.md: https://raw.githubusercontent.com/googleapis/mcp-toolbox/main/docs/en/integrations/bigquery/source.md (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/mcp/bigquery/ --- # One MCP server for Collibra: Collibra MCP Server (chip), from Collibra MCP SERVERS FOR A DATA TOOL, read 10 Nov 2026 **The short answer.** One MCP server for Collibra is listed: Collibra MCP Server (chip), official, published by Collibra and read on 10 Nov 2026. It runs remote and local and names 49 tools; the version read is V0.0.67. Last release: 27 Oct 2026. Auth: Collibra username/password: server-wide credentials over stdio (mcp.yaml or env vars), or client-supplied Basic Auth headers per request over HTTP. - **Data tool:** [Collibra](/mcp/collibra/) - **Publisher:** Collibra - **Kind:** Official - **Repository or docs:** [github.com/collibra/chip](https://github.com/collibra/chip) - **Version or commit:** V0.0.67 - **Last release:** 27 Oct 2026 - **Transport:** Remote and local - **Auth:** Collibra username/password: server-wide credentials over stdio (mcp.yaml or env vars), or client-supplied Basic Auth headers per request over HTTP; tools respect the connecting user's DGC scopes. - **Access:** Configurable - **Tools listed:** 49 named in the evidence - **Tool definitions read at:** [github.com/collibra/chip/blob/main/README.md](https://github.com/collibra/chip/blob/main/README.md) - **Context weight:** 2,914 tokens for 41 tool definitions (41 with a description), measured from the README read - **Status:** Exists on 10 Nov 2026 as collibra/chip; v0.0.67 released 27 Oct 2026 per GitHub releases. Collibra's product-docs page for it returned a Cloudflare block, so only the repo README was read. - **Read:** 10 Nov 2026 ## Collibra MCP Server (chip) Collibra's server covers lineage with transformation logic, business terms and column semantics, classification, stewardship edits and data quality, under the connecting user's permissions. `discover_business_glossary`, `discover_data_assets`, `get_assessment`, `get_asset_details`, `get_business_term_data`, `get_column_semantics`, `get_data_quality_rule`, `get_data_quality_rule_results`, `validate_data_quality_rule`, `list_data_quality_rule_templates`, `get_data_quality_rule_template`, `find_data_quality_rules`, `generate_data_quality_rule_sql`, `get_lineage_downstream`, `get_lineage_entity`, `get_lineage_transformation`, `get_lineage_upstream`, `get_measure_data`, `get_table_semantics`, `get_user_id_by_name`, `list_asset_types`, `list_data_contract`, `prepare_create_asset`, `pull_data_contract_manifest`, `search_asset_keyword`, `search_catalog_columns`, `search_data_class`, `search_data_classification_match`, `search_lineage_entities`, `search_lineage_transformations`, `add_data_classification_match`, `create_assessment`, `create_asset`, `create_data_quality_job`, `create_data_quality_rule`, `deploy_data_quality_rule_template`, `dq_cancel_job_run`, `dq_delete_job`, `dq_delete_job_run`, `dq_update_job`, `dq_get_job`, `dq_get_job_run`, `dq_search_jobs`, `dq_search_job_runs`, `edit_assessment`, `edit_asset`, `init_data_contract`, `push_data_contract_manifest`, `remove_data_classification_match` - None found on 10 Nov 2026: A README statement that the lineage tools return column-level edges or BI-field lineage explicitly; the tool list names entities and transformations only. - None found on 10 Nov 2026: A list of warehouses, BI tools or transformation tools supported by the Collibra MCP README (coverage depends on what is catalogued in the Collibra instance). - None found on 10 Nov 2026: Any statement on collection cadence or freshness in the Collibra MCP README (the server queries a live Collibra instance). ## Read next - [Collibra MCP Server (chip), the full entry](/servers/collibra-mcp/): 23 facts - [Alation AI Agent SDK MCP server and Collibra MCP Server (chip), side by side](/compare/alation-mcp-vs-collibra-mcp/) - [Collibra MCP Server (chip) and Euno MCP, side by side](/compare/collibra-mcp-vs-euno-mcp/) - [Catalogs and metadata](/categories/catalogs/) ## Sources - github.com/collibra/chip/blob/main/README.md: https://github.com/collibra/chip/blob/main/README.md (read 10 Nov 2026) - github.com/collibra/chip/releases: https://github.com/collibra/chip/releases (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/mcp/collibra/ --- # One MCP server for Cube: Cube MCP server, from Cube MCP SERVERS FOR A DATA TOOL, read 10 Nov 2026 **The short answer.** One MCP server for Cube is listed: Cube MCP server, official, published by Cube and read on 10 Nov 2026. It runs remote (hosted) and names 30 tools; the version read is Hosted service (no versioned release). Last release: none found on 10 Nov 2026. - **Data tool:** [Cube](/mcp/cube/) - **Publisher:** Cube - **Kind:** Official - **Repository or docs:** [docs.cube.dev/docs/integrations/mcp-server](https://docs.cube.dev/docs/integrations/mcp-server) - **Version or commit:** Hosted service (no versioned release) - **Last release:** none found on 10 Nov 2026 - **Transport:** Remote (hosted) - **Auth:** OAuth Authorization Code + PKCE (scope mcp-agent-access, public client cube-mcp-client), or a Cube API key exchanged for a token (client_id cube-api-key). - **Access:** Read and write - **Tools listed:** 30 named in the evidence - **Tool definitions read at:** [docs.cube.dev/docs/integrations/mcp-server.md](https://docs.cube.dev/docs/integrations/mcp-server.md) - **Context weight:** 1,093 tokens for 30 tool definitions (30 with a description), measured from the docs page read - **Status:** Exists on 10 Nov 2026 as a Cube-hosted endpoint cubecloud.dev/mcp (or your own console domain). The page says it exposes 30 tools; the tables list 28 names, all captured here. - **Read:** 10 Nov 2026 ## Cube MCP server Cube's hosted server searches the semantic layer, runs SQL with row-level security as the signed-in user and edits the semantic model on a branch. `listDeployments`, `chat`, `loadQueryResults`, `searchDataModel`, `runQuery`, `readWorkbook`, `createWorkbook`, `createReport`, `readReport`, `updateReport`, `deleteReport`, `manageFolders`, `updateDashboard`, `publishDashboard`, `listDataModelFiles`, `readDataModelFile`, `startDataModelEdit`, `writeDataModelFile`, `deleteDataModelFile`, `getDataModelChanges`, `getBranchDiff`, `getDeploymentEnv`, `getBranchState`, `commitDataModelChanges`, `commitToNewBranch`, `mergeToDefaultBranch`, `discardDataModelChanges`, `switchUserBranch`, `getPreAggregationStatus`, `buildPreAggregation` - None found on 10 Nov 2026: A lineage tool in the Cube MCP tool tables (getBranchDiff and getDataModelChanges compare model versions, not data lineage). ## Read next - [Cube MCP server, the full entry](/servers/cube-mcp/): 19 facts - [Cube MCP server and dbt MCP Server, side by side](/compare/cube-mcp-vs-dbt-mcp/) - [BI and analytics](/categories/bi/) ## Sources - docs.cube.dev/docs/integrations/mcp-server.md: https://docs.cube.dev/docs/integrations/mcp-server.md (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/mcp/cube/ --- # 2 MCP servers for Databricks, read from their own sources MCP SERVERS FOR A DATA TOOL, read 10 Nov 2026 **The short answer.** 2 MCP servers for Databricks are listed, each read from its own source on 10 Nov 2026: Databricks Labs MCP servers (official, labs org, remote and local); Databricks managed MCP servers (official, remote (hosted)). Between them they name 7 tools. The table puts the same fields side by side. | Field | Databricks Labs MCP servers | Databricks managed MCP servers | | --- | --- | --- | | Data tool | [Databricks](/mcp/databricks/) | [Databricks](/mcp/databricks/) | | Publisher | Databricks Labs | Databricks | | Kind | Official, Labs org | Official | | Repository or docs | [github.com/databrickslabs/mcp](https://github.com/databrickslabs/mcp) | [docs.databricks.com/aws/en/generative-ai/mcp/managed-mcp](https://docs.databricks.com/aws/en/generative-ai/mcp/managed-mcp) | | Version or commit | none found on 10 Nov 2026 | Docs page last updated Nov 9, 2026 | | Last release | none found on 10 Nov 2026 | none found on 10 Nov 2026 | | Transport | Remote and local | Remote (hosted) | | Auth | Databricks CLI-style credentials; Unity Catalog permissions. | Databricks workspace sign-in; caller's Unity Catalog permissions; Unity Gateway grants and policies. | | Access | Not stated in the source read | Configurable | | Tools listed | 3 named in the evidence | 4 named in the evidence | | Tool definitions read at | [raw.githubusercontent.com/databrickslabs/mcp/master/README.md](https://raw.githubusercontent.com/databrickslabs/mcp/master/README.md) | [docs.databricks.com/aws/en/generative-ai/mcp/managed-mcp](https://docs.databricks.com/aws/en/generative-ai/mcp/managed-mcp) | | Context weight | Not measured: README says the tool list is inferred at startup from the Unity Catalog functions, vector search indexes and Genie spaces the user configures, so no fixed per-tool list or descriptions are published. | 94 tokens for 4 tool definitions (4 with a description), measured from the docs page read | | Status | Exists; README marks Unity Catalog server deprecated in favor of managed MCP servers; last push 26 Oct 2026 (GitHub API); no tagged release read. | Documented by Databricks as Databricks-provided MCPs (read 10 Nov 2026); no versioned release stream found. | | Read | 10 Nov 2026 | 10 Nov 2026 | 2 servers for Databricks, the same fields in the same order, read 10 Nov 2026. ## Databricks Labs MCP servers The Labs server exposes Unity Catalog functions, vector search indexes and Genie spaces as tools; its README marks it deprecated in favor of the managed servers. `UC functions (one tool per function)`, `vector search index tools`, `Genie space tools` - None found on 10 Nov 2026: A lineage tool in the README supported-tools list. ## Databricks managed MCP servers Databricks-provided MCP endpoints run under the caller's Unity Catalog grants and include Genie spaces for questions over a governed semantic layer. `system.ai.genie_one_mcp`, `system.ai.dbsql`, `system.ai.web_search`, `system.ai.sandbox` - None found on 10 Nov 2026: A lineage or column-lineage MCP on the Databricks-provided MCPs page (list read). ## Read next - [Databricks Labs MCP servers, the full entry](/servers/databrickslabs-mcp/): 11 facts - [Databricks managed MCP servers, the full entry](/servers/databricks-managed-mcp/): 12 facts - [Databricks Labs MCP servers and Databricks managed MCP servers, side by side](/compare/databrickslabs-mcp-vs-databricks-managed-mcp/) - [Warehouses](/categories/warehouses/) ## Sources - raw.githubusercontent.com/databrickslabs/mcp/master/README.md: https://raw.githubusercontent.com/databrickslabs/mcp/master/README.md (read 10 Nov 2026) - docs.databricks.com/aws/en/generative-ai/mcp/managed-mcp: https://docs.databricks.com/aws/en/generative-ai/mcp/managed-mcp (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/mcp/databricks/ --- # One MCP server for DataHub: DataHub MCP Server, from Acryl Data MCP SERVERS FOR A DATA TOOL, read 10 Nov 2026 **The short answer.** One MCP server for DataHub is listed: DataHub MCP Server, official, published by Acryl Data and read on 10 Nov 2026. It runs remote and local and names 22 tools; the version read is 0.8.0. Last release: 21 Oct 2026. Auth: Stdio: DATAHUB_GMS_URL + DATAHUB_GMS_TOKEN (or ~/.datahubenv). - **Data tool:** [DataHub](/mcp/datahub/) - **Publisher:** Acryl Data (DataHub) - **Kind:** Official - **Repository or docs:** [github.com/acryldata/mcp-server-datahub](https://github.com/acryldata/mcp-server-datahub) - **Version or commit:** 0.8.0 - **Last release:** 21 Oct 2026 - **Transport:** Remote and local - **Auth:** Stdio: DATAHUB_GMS_URL + DATAHUB_GMS_TOKEN (or ~/.datahubenv); HTTP: per-request DataHub personal access token as Authorization Bearer, no shared server token. - **Access:** Configurable - **Tools listed:** 22 named in the evidence - **Tool definitions read at:** [github.com/acryldata/mcp-server-datahub/blob/main/README.md](https://github.com/acryldata/mcp-server-datahub/blob/main/README.md) - **Context weight:** 567 tokens for 22 tool definitions (22 with a description), measured from the README read - **Status:** Exists under this name on 10 Nov 2026; v0.8.0 released 21 Oct 2026 per GitHub releases and PyPI. - **Read:** 10 Nov 2026 ## DataHub MCP Server DataHub's server returns column-level lineage with paths and SQL transformations, plus glossary, domain and owner discovery, with per-user tokens in HTTP mode. `search`, `get_lineage`, `get_dataset_queries`, `get_entities`, `get_aspect_history`, `list_schema_fields`, `get_lineage_paths_between`, `add_tags`, `remove_tags`, `add_terms`, `remove_terms`, `add_owners`, `remove_owners`, `set_domains`, `remove_domains`, `update_description`, `add_structured_properties`, `remove_structured_properties`, `get_me`, `search_documents`, `grep_documents`, `save_document` - None found on 10 Nov 2026: Any statement in the DataHub MCP README that tags, terms or classifications propagate along lineage when read through the MCP tools. - None found on 10 Nov 2026: Any freshness or ingestion-cadence statement in the DataHub MCP README (the server reads whatever DataHub has ingested). The README only describes caching of verified tokens for five minutes. ## Read next - [DataHub MCP Server, the full entry](/servers/datahub-mcp/): 19 facts - [Atlan MCP Server and DataHub MCP Server, side by side](/compare/atlan-mcp-vs-datahub-mcp/) - [DataHub MCP Server and OpenMetadata MCP Server, side by side](/compare/datahub-mcp-vs-openmetadata-mcp/) - [DataHub MCP Server and Euno MCP, side by side](/compare/datahub-mcp-vs-euno-mcp/) - [Catalogs and metadata](/categories/catalogs/) ## Sources - raw.githubusercontent.com/acryldata/mcp-server-datahub/main/README.md: https://raw.githubusercontent.com/acryldata/mcp-server-datahub/main/README.md (read 10 Nov 2026) - pypi.org/pypi/mcp-server-datahub/json: https://pypi.org/pypi/mcp-server-datahub/json (read 10 Nov 2026) - github.com/acryldata/mcp-server-datahub/releases/tag/v0.8.0: https://github.com/acryldata/mcp-server-datahub/releases/tag/v0.8.0 (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/mcp/datahub/ --- # One MCP server for dbt: dbt MCP Server, from dbt Labs MCP SERVERS FOR A DATA TOOL, read 10 Nov 2026 **The short answer.** One MCP server for dbt is listed: dbt MCP Server, official, published by dbt Labs and read on 10 Nov 2026. It runs remote and local and names 52 tools; the version read is 2.10.1. Last release: 3 Nov 2026. Auth: dbt service/personal token and environment IDs for platform tools. - **Data tool:** [dbt](/mcp/dbt/) - **Publisher:** dbt Labs - **Kind:** Official - **Repository or docs:** [github.com/dbt-labs/dbt-mcp](https://github.com/dbt-labs/dbt-mcp) - **Version or commit:** 2.10.1 - **Last release:** 3 Nov 2026 - **Transport:** Remote and local - **Auth:** dbt service/personal token and environment IDs for platform tools; OAuth or token per docs; local dbt project for CLI tools. - **Access:** Configurable - **Tools listed:** 52 named in the evidence - **Tool definitions read at:** [raw.githubusercontent.com/dbt-labs/dbt-mcp/main/README.md](https://raw.githubusercontent.com/dbt-labs/dbt-mcp/main/README.md) - **Context weight:** 1,061 tokens for 63 tool definitions (63 with a description), measured from the README read - **Status:** Exists, v2.10.1 on 3 Nov 2026 (PyPI and GitHub). - **Read:** 10 Nov 2026 ## dbt MCP Server dbt Labs' server combines the dbt CLI, the Discovery API with model and column lineage, the Semantic Layer and SQL tools on dbt Platform. `execute_sql`, `text_to_sql`, `get_dimension_values`, `get_dimensions`, `get_entities`, `get_metrics_compiled_sql`, `list_metrics`, `list_saved_queries`, `query_metrics`, `get_all_macros`, `get_all_models`, `get_all_sources`, `get_exposures`, `get_lineage`, `get_mart_models`, `get_model_health`, `get_model_performance`, `get_node_details`, `get_related_models`, `search_public_mesh_models`, `build`, `clone`, `compile`, `docs`, `get_lineage_dev`, `get_node_details_dev`, `list`, `parse`, `run`, `show`, `test`, `cancel_job_run`, `get_job_details`, `get_job_run_artifacts`, `get_job_run_details`, `get_job_run_error`, `list_job_run_artifacts`, `list_jobs`, `list_jobs_runs`, `list_projects`, `retry_job_run`, `trigger_job_run`, `generate_model_yaml`, `generate_source`, `generate_staging_model`, `fusion.compile_sql`, `fusion.get_column_lineage`, `get_column_lineage`, `get_product_doc_pages`, `search_product_docs`, `get_mcp_server_branch`, `get_mcp_server_version` ## Read next - [dbt MCP Server, the full entry](/servers/dbt-mcp/): 14 facts - [Cube MCP server and dbt MCP Server, side by side](/compare/cube-mcp-vs-dbt-mcp/) - [Transformation](/categories/transformation/) ## Sources - github.com/dbt-labs/dbt-mcp/releases: https://github.com/dbt-labs/dbt-mcp/releases (read 10 Nov 2026) - docs.getdbt.com/docs/dbt-ai/about-mcp: https://docs.getdbt.com/docs/dbt-ai/about-mcp (read 10 Nov 2026) - raw.githubusercontent.com/dbt-labs/dbt-mcp/main/README.md: https://raw.githubusercontent.com/dbt-labs/dbt-mcp/main/README.md (read 10 Nov 2026) - raw.githubusercontent.com/dbt-labs/dbt-mcp/main/.env.example: https://raw.githubusercontent.com/dbt-labs/dbt-mcp/main/.env.example (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/mcp/dbt/ --- # One MCP server for Euno: Euno MCP, from Euno MCP SERVERS FOR A DATA TOOL, read 10 Nov 2026 **The short answer.** One MCP server for Euno is listed: Euno MCP, official, published by Euno and read on 10 Nov 2026. It runs remote (hosted) and names 56 tools; the version read is Hosted service at api.app.euno.ai/mcp, no version published (docs read 10 Nov 2026). Last release: none found on 10 Nov 2026. - **Data tool:** [Euno](/mcp/euno/) - **Publisher:** Euno - **Kind:** Official - **Repository or docs:** [docs.euno.ai/ai/mcp-setup](https://docs.euno.ai/ai/mcp-setup?utm_source=datamcpservers.com&utm_medium=referral&utm_campaign=tool) - **Version or commit:** Hosted service at api.app.euno.ai/mcp, no version published (docs read 10 Nov 2026) - **Last release:** none found on 10 Nov 2026 - **Transport:** Remote (hosted) - **Auth:** OAuth sign-in (default) or API key and headers; active persona scopes data and actions. - **Access:** Configurable - **Tools listed:** 56 named in the evidence - **Tool definitions read at:** [docs.euno.ai/ai/mcp-setup.md](https://docs.euno.ai/ai/mcp-setup.md?utm_source=datamcpservers.com&utm_medium=referral&utm_campaign=tool) - **Context weight:** 760 tokens for 56 tool definitions (42 with a description), measured from the docs page read - **Status:** Exists on 10 Nov 2026; documented in docs.euno.ai MCP Setup Guide; tool names and descriptions published, parameter schemas not published. - **Read:** 10 Nov 2026 ## Euno MCP Euno's hosted OAuth endpoint serves persona-scoped lineage, impact analysis, EQL search and catalog tools over its metadata graph, with no warehouse query tool. `load_account_settings`, `list_personas`, `current_persona`, `switch_persona`, `load_skill`, `search_properties`, `fetch_single_resource`, `find_resource_by_name`, `find_resources_for_topic`, `get_upstream_lineage`, `resource_impact_analysis`, `query_to_eql`, `search_resources_eql`, `count_resources_eql`, `multi_count_resources_eql`, `pivot_resources_eql`, `search_resource_audit_log`, `sensitive_data_incident_response`, `documentation_search`, `documentation_get_full_document`, `documentation_get_surrounding_context`, `list_metadata_tags`, `list_metadata_tag_domains`, `preview_metadata_tag_plan`, `get_context_benchmark_overview`, `get_context_benchmark_questions`, `list_context_benchmark_runs`, `get_context_benchmark_run`, `get_context_benchmark_improvement_report`, `get_context_benchmark_question_result`, `get_context_benchmark_suggestion_application`, `get_context_benchmark_schedule`, `create_context_benchmark_question`, `update_context_benchmark_question`, `delete_context_benchmark_question`, `start_context_benchmark_run`, `stop_context_benchmark_run`, `update_context_benchmark_schedule`, `prepare_context_benchmark_suggestion_application`, `decide_context_benchmark_suggestion_application`, `undo_context_benchmark_suggestion`, `reapply_context_benchmark_suggestion`, `fetch_glossary_term`, `find_glossary_term`, `create_glossary_term`, `update_glossary_term`, `delete_glossary_term`, `fetch_metadata_tag`, `create_metadata_tag`, `update_metadata_tag`, `delete_metadata_tag`, `fetch_workflow`, `find_workflow`, `create_workflow`, `update_workflow`, `delete_workflow` - None found on 10 Nov 2026: A version number or dated release list for the Euno MCP server; the docs describe a hosted service, and the separate Euno SDK is v0.5. **What Euno's server does not do.** Euno's server gives an agent governed context about the data; it does not run the query, which is the job of the warehouse's own server. ## Read next - [Euno MCP, the full entry](/servers/euno-mcp/): 21 facts - [Atlan MCP Server and Euno MCP, side by side](/compare/atlan-mcp-vs-euno-mcp/) - [DataHub MCP Server and Euno MCP, side by side](/compare/datahub-mcp-vs-euno-mcp/) - [Collibra MCP Server (chip) and Euno MCP, side by side](/compare/collibra-mcp-vs-euno-mcp/) - [Catalogs and metadata](/categories/catalogs/) ## Sources - docs.euno.ai/ai/mcp-setup.md: https://docs.euno.ai/ai/mcp-setup.md (read 10 Nov 2026) - docs.euno.ai/ai/mcp-setup/mcp-bridge-setup.md: https://docs.euno.ai/ai/mcp-setup/mcp-bridge-setup.md (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/mcp/euno/ --- # One MCP server for Fivetran: Fivetran MCP Server, from Fivetran MCP SERVERS FOR A DATA TOOL, read 10 Nov 2026 **The short answer.** One MCP server for Fivetran is listed: Fivetran MCP Server, official, published by Fivetran and read on 10 Nov 2026. It runs remote and local and names 53 tools; the version read is V0.5.1. Last release: 6 Oct 2026. Auth: Fivetran API key and API secret (FIVETRAN_API_KEY, FIVETRAN_API_SECRET) from the Fivetran dashboard. - **Data tool:** [Fivetran](/mcp/fivetran/) - **Publisher:** Fivetran - **Kind:** Official - **Repository or docs:** [github.com/fivetran/fivetran-mcp](https://github.com/fivetran/fivetran-mcp) - **Version or commit:** V0.5.1 - **Last release:** 6 Oct 2026 - **Transport:** Remote and local - **Auth:** Fivetran API key and API secret (FIVETRAN_API_KEY, FIVETRAN_API_SECRET) from the Fivetran dashboard; one shared credential, not per-user OAuth. Streamable HTTP mode forwards headers; OAuth resource-server support is scaffolding only. - **Access:** Configurable - **Tools listed:** 53 named in the evidence - **Tool definitions read at:** [raw.githubusercontent.com/fivetran/fivetran-mcp/main/open-api-definitions/AVAILABLE_ACTIONS.md](https://raw.githubusercontent.com/fivetran/fivetran-mcp/main/open-api-definitions/AVAILABLE_ACTIONS.md) - **Context weight:** 299 tokens for 53 tool definitions (2 with a description), measured from the README read - **Status:** Exists on 10 Nov 2026 as fivetran/fivetran-mcp, installed from git with uvx (the PyPI names fivetran-mcp and mcp-fivetran belong to unrelated projects). Tool set is generated from the OpenAPI manifest, so it mirrors the REST API. - **Read:** 10 Nov 2026 ## Fivetran MCP Server Fivetran's server mirrors its REST API behind a read-by-default scope with a deny-list, so an agent can inspect connections, schemas and sync status. `list_endpoints`, `get_schema`, `account_read`, `connections_delete`, `connections_read`, `connections_write`, `connector_sdk_delete`, `connector_sdk_read`, `connector_sdk_write`, `destinations_delete`, `destinations_read`, `destinations_write`, `external_logging_delete`, `external_logging_read`, `external_logging_write`, `external_secrets_managers_delete`, `external_secrets_managers_read`, `external_secrets_managers_write`, `external_secrets_managers_entities_read`, `groups_delete`, `groups_read`, `groups_write`, `hvr_write`, `hybrid_deployment_agents_delete`, `hybrid_deployment_agents_read`, `hybrid_deployment_agents_write`, `metadata_read`, `private_links_delete`, `private_links_read`, `private_links_write`, `proxy_delete`, `proxy_read`, `proxy_write`, `public_read`, `roles_read`, `system_keys_delete`, `system_keys_read`, `system_keys_write`, `teams_delete`, `teams_read`, `teams_write`, `transformation_projects_delete`, `transformation_projects_read`, `transformation_projects_write`, `transformations_delete`, `transformations_read`, `transformations_write`, `users_delete`, `users_read`, `users_write`, `webhooks_delete`, `webhooks_read`, `webhooks_write` - None found on 10 Nov 2026: Any lineage path in the OpenAPI definition the server loads (searched fivetran-open-api-definition.json for 'lineage' in endpoint paths). - None found on 10 Nov 2026: Glossary, ownership or classification tools in the Fivetran MCP README or AVAILABLE_ACTIONS.md. ## Read next - [Fivetran MCP Server, the full entry](/servers/fivetran-mcp/): 20 facts - [Airbyte MCP (hosted) and Fivetran MCP Server, side by side](/compare/airbyte-mcp-vs-fivetran-mcp/) - [Ingestion connectors](/categories/ingestion/) ## Sources - raw.githubusercontent.com/fivetran/fivetran-mcp/main/README.md: https://raw.githubusercontent.com/fivetran/fivetran-mcp/main/README.md (read 10 Nov 2026) - github.com/fivetran/fivetran-mcp/releases: https://github.com/fivetran/fivetran-mcp/releases (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/mcp/fivetran/ --- # One MCP server for Hex: Hex MCP server, from Hex MCP SERVERS FOR A DATA TOOL, read 10 Nov 2026 **The short answer.** One MCP server for Hex is listed: Hex MCP server, official, published by Hex and read on 10 Nov 2026. It runs remote (hosted) and names 18 tools; the version read is Hosted service (no versioned release). Last release: none found on 10 Nov 2026. - **Data tool:** [Hex](/mcp/hex/) - **Publisher:** Hex - **Kind:** Official - **Repository or docs:** [learn.hex.tech/docs/api-integrations/mcp-server](https://learn.hex.tech/docs/api-integrations/mcp-server) - **Version or commit:** Hosted service (no versioned release) - **Last release:** none found on 10 Nov 2026 - **Transport:** Remote (hosted) - **Auth:** OAuth authorization to the user's Hex workspace (server URL app.hex.tech/mcp, or the customer's single-tenant, EU or HIPAA domain); requires Explorer role or higher. - **Access:** Read and write - **Tools listed:** 18 named in the evidence - **Tool definitions read at:** [learn.hex.tech/docs/api-integrations/mcp-server](https://learn.hex.tech/docs/api-integrations/mcp-server) - **Context weight:** 252 tokens for 18 tool definitions (18 with a description), measured from the docs page read - **Status:** Exists on 10 Nov 2026 as a Hex-hosted MCP server; tool list read from the Hex docs page (knowledge tools for Explorers, project editing tools for Editors). - **Read:** 10 Nov 2026 ## Hex MCP server Hex's hosted server runs Threads (agent analyses over the user's data connections) and notebook editing under workspace roles and project permissions. `search_projects`, `create_thread`, `get_thread`, `continue_thread`, `get_me`, `create_project`, `get_project`, `list_cells`, `get_cell`, `create_cell`, `update_cell`, `delete_cell`, `run_cell`, `run_notebook`, `get_run`, `get_cell_output`, `get_cell_image`, `list_data_connections` - None found on 10 Nov 2026: A lineage tool on the Hex MCP server page (cell-level reads exist, but no column or upstream-table lineage tool is listed). - None found on 10 Nov 2026: Scheduled metadata collection on the Hex MCP page. ## Read next - [Hex MCP server, the full entry](/servers/hex-mcp/): 18 facts - [Hex MCP server and Omni MCP Server, side by side](/compare/hex-mcp-vs-omni-mcp/) - [BI and analytics](/categories/bi/) ## Sources - learn.hex.tech/docs/api-integrations/mcp-server: https://learn.hex.tech/docs/api-integrations/mcp-server (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/mcp/hex/ --- # One MCP server for Looker: Looker tools in MCP Toolbox for Databases, from Google MCP SERVERS FOR A DATA TOOL, read 10 Nov 2026 **The short answer.** One MCP server for Looker is listed: Looker tools in MCP Toolbox for Databases, official, published by Google and read on 10 Nov 2026. It runs remote and local and names 55 tools; the version read is V1.16.0. Last release: 5 Nov 2026. 23 sourced facts follow, each with its read date. - **Data tool:** [Looker](/mcp/looker/) - **Publisher:** Google (googleapis/mcp-toolbox) - **Kind:** Official - **Repository or docs:** [github.com/googleapis/mcp-toolbox](https://github.com/googleapis/mcp-toolbox) - **Version or commit:** V1.16.0 - **Last release:** 5 Nov 2026 - **Transport:** Remote and local - **Auth:** Looker API client_id and client_secret in the Looker source config, or use_client_oauth to pass the caller's OAuth token (default false). - **Access:** Configurable - **Tools listed:** 55 named in the evidence - **Tool definitions read at:** [mcp-toolbox.dev/llms.txt](https://mcp-toolbox.dev/llms.txt) - **Context weight:** 1,475 tokens for 55 tool definitions (55 with a description), measured from the docs index read - **Status:** Exists on 10 Nov 2026 as the looker source type in MCP Toolbox for Databases; the repo URL github.com/googleapis/genai-toolbox now titles itself googleapis/mcp-toolbox. 55 looker-* tools are listed in the Toolbox llms.txt index; schema file holds the one-line descriptions only. - **Read:** 10 Nov 2026 ## Looker tools in MCP Toolbox for Databases Google's Toolbox exposes Looker models, explores, dimensions, measures and LookML project files, plus instance health and usage tools. `looker-add-dashboard-element`, `looker-add-dashboard-filter`, `looker-conversational-analytics`, `looker-create-agent`, `looker-create-dashboard-layout`, `looker-create-git-branch`, `looker-create-project-directory`, `looker-create-project-file`, `looker-delete-agent`, `looker-delete-git-branch`, `looker-delete-project-directory`, `looker-delete-project-file`, `looker-dev-mode`, `looker-generate-embed-url`, `looker-get-agent`, `looker-get-connection-databases`, `looker-get-connection-schemas`, `looker-get-connection-table-columns`, `looker-get-connection-tables`, `looker-get-connections`, `looker-get-dashboard`, `looker-get-dashboards`, `looker-get-dimensions`, `looker-get-explore`, `looker-get-explores`, `looker-get-field-value-suggestions`, `looker-get-filters`, `looker-get-git-branch`, `looker-get-looks`, `looker-get-measures`, `looker-get-models`, `looker-get-parameters`, `looker-get-project-directories`, `looker-get-project-file`, `looker-get-project-files`, `looker-get-projects`, `looker-health-analyze`, `looker-health-pulse`, `looker-health-vacuum`, `looker-import-dashboard-from-lookml`, `looker-list-agents`, `looker-list-git-branches`, `looker-make-dashboard`, `looker-make-look`, `looker-query`, `looker-query-sql`, `looker-query-url`, `looker-run-dashboard`, `looker-run-look`, `looker-switch-git-branch`, `looker-update-agent`, `looker-update-dashboard-element`, `looker-update-dashboard-layout-component`, `looker-update-project-file`, `looker-validate-project` - None found on 10 Nov 2026: A lineage tool among the 55 looker-* tools in the Toolbox llms.txt index (no column-lineage or upstream-table tool is named). - None found on 10 Nov 2026: Scheduled or event-driven metadata collection for Looker in the Toolbox docs; tools call the Looker API per request. ## Read next - [Looker tools in MCP Toolbox for Databases, the full entry](/servers/looker-mcp-toolbox/): 23 facts - [Looker tools in MCP Toolbox for Databases and Tableau MCP, side by side](/compare/looker-mcp-toolbox-vs-tableau-mcp/) - [BI and analytics](/categories/bi/) ## Sources - raw.githubusercontent.com/googleapis/genai-toolbox/main/README.md: https://raw.githubusercontent.com/googleapis/genai-toolbox/main/README.md (read 10 Nov 2026) - github.com/googleapis/mcp-toolbox/releases: https://github.com/googleapis/mcp-toolbox/releases (read 10 Nov 2026) - mcp-toolbox.dev/integrations/looker/source: https://mcp-toolbox.dev/integrations/looker/source/ (read 10 Nov 2026) - mcp-toolbox.dev/llms.txt: https://mcp-toolbox.dev/llms.txt (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/mcp/looker/ --- # One MCP server for Monte Carlo: Monte Carlo MCP Server, from Monte Carlo MCP SERVERS FOR A DATA TOOL, read 10 Nov 2026 **The short answer.** One MCP server for Monte Carlo is listed: Monte Carlo MCP Server, official, published by Monte Carlo and read on 10 Nov 2026. It runs remote (hosted) and names 15 tools; the version read is Hosted at mcp.getmontecarlo.com/mcp, no version published. Last release: none found on 10 Nov 2026. - **Data tool:** [Monte Carlo](/mcp/monte-carlo/) - **Publisher:** Monte Carlo - **Kind:** Official - **Repository or docs:** [docs.getmontecarlo.com/docs/mcp-server](https://docs.getmontecarlo.com/docs/mcp-server) - **Version or commit:** Hosted at mcp.getmontecarlo.com/mcp, no version published - **Last release:** none found on 10 Nov 2026 - **Transport:** Remote (hosted) - **Auth:** OAuth 2.1 with Dynamic Client Registration (user identity, Editor role or above), or MCP-server-type API key id:secret as Bearer token via mcp-remote. - **Access:** Configurable - **Tools listed:** 15 named in the evidence - **Tool definitions read at:** none found on 10 Nov 2026 - **Context weight:** Not measured: no published tool-definition text was read on 10 Nov 2026 - **Status:** Exists on 10 Nov 2026 as a hosted first-party server; tool names are partial (those named in the docs), full tool definitions were not published as text, so no schema file was saved. - **Read:** 10 Nov 2026 ## Monte Carlo MCP Server Monte Carlo's hosted observability server uses per-user OAuth or scoped keys, a per-request read-only header and tool allow-lists. `search`, `get_table`, `get_asset_lineage`, `get_alerts`, `alert_assessment`, `update_alert`, `get_monitors`, `create_or_update_table_monitor`, `get_use_cases`, `get_use_case_tables`, `get_use_case_table_summary`, `get_agent_metadata`, `get_agent_conversation`, `get_agent_trace`, `create_agent_metric_monitor` - None found on 10 Nov 2026: Documentation that get_asset_lineage returns column-level lineage, transformation SQL, BI fields or metrics; the MCP pages name only asset lineage. - None found on 10 Nov 2026: Glossary terms, definitions or classification tools in the Monte Carlo MCP server docs; the tools are alerts, monitors, assets, lineage and data operations. - None found on 10 Nov 2026: A list of warehouses and BI systems on the MCP pages (the server fronts the Monte Carlo account's monitored assets). - None found on 10 Nov 2026: A freshness statement on the MCP pages (the server reads live alert and monitor state from the Monte Carlo account). ## Read next - [Monte Carlo MCP Server, the full entry](/servers/monte-carlo-mcp/): 18 facts - [Monte Carlo MCP Server and SYNQ MCP (Coalesce Quality), side by side](/compare/monte-carlo-mcp-vs-synq-mcp/) - [Monte Carlo MCP Server and Sifflet MCP Server, side by side](/compare/monte-carlo-mcp-vs-sifflet-mcp/) - [Data observability](/categories/observability/) ## Sources - docs.getmontecarlo.com/docs/mcp-server: https://docs.getmontecarlo.com/docs/mcp-server (read 10 Nov 2026) - docs.getmontecarlo.com/docs/mcp-server-reference-and-operations: https://docs.getmontecarlo.com/docs/mcp-server-reference-and-operations (read 10 Nov 2026) - docs.getmontecarlo.com/docs/ai-security: https://docs.getmontecarlo.com/docs/ai-security (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/mcp/monte-carlo/ --- # One MCP server for Omni: Omni MCP Server, from Omni Analytics MCP SERVERS FOR A DATA TOOL, read 10 Nov 2026 **The short answer.** One MCP server for Omni is listed: Omni MCP Server, official, published by Omni Analytics and read on 10 Nov 2026. It runs remote (hosted) and names 26 tools; the version read is Hosted service (no versioned release). Last release: none found on 10 Nov 2026. - **Data tool:** [Omni](/mcp/omni/) - **Publisher:** Omni Analytics - **Kind:** Official - **Repository or docs:** [docs.omni.co/ai/mcp](https://docs.omni.co/ai/mcp) - **Version or commit:** Hosted service (no versioned release) - **Last release:** none found on 10 Nov 2026 - **Transport:** Remote (hosted) - **Auth:** OAuth 2.1 with PKCE (creates an MCP OAuth PAT for the user), or an API key (personal token or Organization API key; Organization keys can act as a user with X-MCP-User-ID). - **Access:** Configurable - **Tools listed:** 26 named in the evidence - **Tool definitions read at:** [docs.omni.co/ai/mcp/tools.md](https://docs.omni.co/ai/mcp/tools.md) - **Context weight:** 466 tokens for 27 tool definitions (27 with a description), measured from the docs page read - **Status:** Exists on 10 Nov 2026 as Omni's hosted MCP server (OAuth callback via callbacks.omniapp.co); admins toggle tool capabilities in AI Hub > MCP. The docs page lists 25 tools. - **Read:** 10 Nov 2026 ## Omni MCP Server Omni's hosted server runs as a specific Omni user and exposes models, topics, saved queries and apps, so answers follow the semantic model and the user's permissions. `pickModel`, `pickTopic`, `getData`, `runQuery`, `listLlms`, `askOmni`, `checkStatus`, `cancelJob`, `listBranches`, `listSkills`, `submitFeedback`, `searchContent`, `listDocumentQueries`, `runSavedQuery`, `startDashboardDownload`, `checkDashboardDownload`, `fetchDashboardDownload`, `getApp`, `getAppDraft`, `putApp`, `patchApp`, `publishDraft`, `showApp`, `import-claude-dashboard-from-url`, `get-dashboard-import-job-status`, `queryUsageAnalytics` - None found on 10 Nov 2026: A lineage tool among the Omni MCP tools (no column or upstream-table lineage tool documented). - None found on 10 Nov 2026: Scheduled or event-driven metadata collection in the Omni MCP docs; tools run queries and read content at call time. ## Read next - [Omni MCP Server, the full entry](/servers/omni-mcp/): 19 facts - [Hex MCP server and Omni MCP Server, side by side](/compare/hex-mcp-vs-omni-mcp/) - [BI and analytics](/categories/bi/) ## Sources - docs.omni.co/ai/mcp/tools.md: https://docs.omni.co/ai/mcp/tools.md (read 10 Nov 2026) - docs.omni.co/ai/mcp/authentication.md: https://docs.omni.co/ai/mcp/authentication.md (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/mcp/omni/ --- # One MCP server for OpenMetadata: OpenMetadata MCP Server, from OpenMetadata MCP SERVERS FOR A DATA TOOL, read 10 Nov 2026 **The short answer.** One MCP server for OpenMetadata is listed: OpenMetadata MCP Server, official, published by OpenMetadata and read on 10 Nov 2026. It runs remote (hosted) and names 15 tools; the version read is 2.0.7 (bundled with platform release 2.0.7-release; 2.1.0-rc1 also listed). Last release: 27 Oct 2026. - **Data tool:** [OpenMetadata](/mcp/openmetadata/) - **Publisher:** OpenMetadata - **Kind:** Official - **Repository or docs:** [docs.open-metadata.org/v2.0.x/how-to-guides/mcp.md](https://docs.open-metadata.org/v2.0.x/how-to-guides/mcp.md) - **Version or commit:** 2.0.7 (bundled with platform release 2.0.7-release; 2.1.0-rc1 also listed) - **Last release:** 27 Oct 2026 - **Transport:** Remote (hosted) - **Auth:** OAuth 2.0 via existing SSO (per user), personal access token, or Bot JWT as Authorization Bearer. - **Access:** Read and write - **Tools listed:** 15 named in the evidence - **Tool definitions read at:** [docs.open-metadata.org/v2.0.x/how-to-guides/mcp/reference.md](https://docs.open-metadata.org/v2.0.x/how-to-guides/mcp/reference.md) - **Context weight:** 1,249 tokens for 15 tool definitions (15 with a description), measured from the docs page read - **Status:** Exists on 10 Nov 2026; built into the OpenMetadata server (not a separate package). Tool access is governed by the caller's role and policies; no read-only switch is documented on the pages read. - **Read:** 10 Nov 2026 ## OpenMetadata MCP Server OpenMetadata's built-in server offers per-user OAuth, persona-scoped context, table and optional column lineage, and write tools for lineage, entities and tests. `search_metadata`, `semantic_search`, `company_context`, `get_entity_details`, `get_user_context`, `get_persona_context`, `find_context`, `get_entity_lineage`, `create_lineage`, `root_cause_analysis`, `create_entity`, `describe_entity_type`, `patch_entity`, `get_test_definitions`, `create_test_case` - None found on 10 Nov 2026: Documentation that the MCP lineage tools return transformation SQL or logic. - None found on 10 Nov 2026: Freshness or collection-cadence statement on the OpenMetadata MCP pages. ## Read next - [OpenMetadata MCP Server, the full entry](/servers/openmetadata-mcp/): 18 facts - [DataHub MCP Server and OpenMetadata MCP Server, side by side](/compare/datahub-mcp-vs-openmetadata-mcp/) - [Catalogs and metadata](/categories/catalogs/) ## Sources - docs.open-metadata.org/v2.0.x/how-to-guides/mcp/reference.md: https://docs.open-metadata.org/v2.0.x/how-to-guides/mcp/reference.md (read 10 Nov 2026) - docs.open-metadata.org/v2.0.x/how-to-guides/mcp.md: https://docs.open-metadata.org/v2.0.x/how-to-guides/mcp.md (read 10 Nov 2026) - github.com/open-metadata/OpenMetadata/releases/tag/2.0.7-release: https://github.com/open-metadata/OpenMetadata/releases/tag/2.0.7-release (read 10 Nov 2026) - docs.open-metadata.org/v2.0.x/how-to-guides/mcp/connect.md: https://docs.open-metadata.org/v2.0.x/how-to-guides/mcp/connect.md (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/mcp/openmetadata/ --- # One MCP server for Power BI: Power BI Modeling MCP Server, from Microsoft MCP SERVERS FOR A DATA TOOL, read 10 Nov 2026 **The short answer.** One MCP server for Power BI is listed: Power BI Modeling MCP Server, official, published by Microsoft and read on 10 Nov 2026. It runs remote and local and names 21 tools; the version read is 1.0.1. Last release: 6 Nov 2026. 21 sourced facts follow, each with its read date. - **Data tool:** [Power BI](/mcp/power-bi/) - **Publisher:** Microsoft - **Kind:** Official - **Repository or docs:** [github.com/microsoft/powerbi-modeling-mcp](https://github.com/microsoft/powerbi-modeling-mcp) - **Version or commit:** 1.0.1 - **Last release:** 6 Nov 2026 - **Transport:** Remote and local - **Auth:** Local: Microsoft Entra interactive sign-in (default) or service principal (client secret or certificate). Hosted: Entra ID as the signed-in user. - **Access:** Configurable - **Tools listed:** 21 named in the evidence - **Tool definitions read at:** [raw.githubusercontent.com/microsoft/powerbi-modeling-mcp/main/README.md](https://raw.githubusercontent.com/microsoft/powerbi-modeling-mcp/main/README.md) - **Context weight:** 387 tokens for 21 tool definitions (21 with a description), measured from the README read - **Status:** Exists on 10 Nov 2026. Microsoft documents two Power BI MCP servers: the Authoring server (local GA over stdio from microsoft/powerbi-modeling-mcp, hosted preview at api.fabric.microsoft.com/v1/mcp/powerbi/authoring and Fabric IQ for answering questions over governed data; an earlier consumption MCP remains documented. Hosted authoring tool names are not published, so the tool list is the local server's. - **Read:** 10 Nov 2026 ## Power BI Modeling MCP Server Microsoft's authoring server reads and writes semantic-model objects (tables, columns, DAX measures, relationships, RLS roles, TMDL) under the user's Fabric permissions, with a read-only switch. `connection_operations`, `database_operations`, `transaction_operations`, `model_operations`, `table_operations`, `column_operations`, `measure_operations`, `relationship_operations`, `dax_query_operations`, `trace_operations`, `partition_operations`, `user_hierarchy_operations`, `calculation_group_operations`, `security_role_operations`, `perspective_operations`, `named_expression_operations`, `function_operations`, `culture_operations`, `object_translation_operations`, `calendar_operations`, `query_group_operations` - None found on 10 Nov 2026: A lineage tool in the Power BI Authoring MCP tool table (no upstream-table, column-lineage or report-dependency tool). - None found on 10 Nov 2026: Scheduled metadata collection or lineage refresh in the Power BI Authoring MCP README or Learn page. ## Read next - [Power BI Modeling MCP Server, the full entry](/servers/powerbi-modeling-mcp/): 21 facts - [Power BI Modeling MCP Server and Tableau MCP, side by side](/compare/powerbi-modeling-mcp-vs-tableau-mcp/) - [BI and analytics](/categories/bi/) ## Sources - learn.microsoft.com/en-us/power-bi/developer/mcp/power-bi-authoring-mcp: https://learn.microsoft.com/en-us/power-bi/developer/mcp/power-bi-authoring-mcp (read 10 Nov 2026) - github.com/microsoft/powerbi-modeling-mcp/releases: https://github.com/microsoft/powerbi-modeling-mcp/releases (read 10 Nov 2026) - registry.npmjs.org/@microsoft/powerbi-modeling-mcp: https://registry.npmjs.org/@microsoft/powerbi-modeling-mcp (read 10 Nov 2026) - learn.microsoft.com/en-us/power-bi/developer/mcp/mcp-servers-overview: https://learn.microsoft.com/en-us/power-bi/developer/mcp/mcp-servers-overview (read 10 Nov 2026) - raw.githubusercontent.com/microsoft/powerbi-modeling-mcp/main/README.md: https://raw.githubusercontent.com/microsoft/powerbi-modeling-mcp/main/README.md (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/mcp/power-bi/ --- # One MCP server for Secoda: Secoda MCP, from Secoda MCP SERVERS FOR A DATA TOOL, read 10 Nov 2026 **The short answer.** One MCP server for Secoda is listed: Secoda MCP, official, published by Secoda and read on 10 Nov 2026. It runs remote and local and names 8 tools; the version read is No releases; main branch last commit 26 Jan 2026. Last release: none found on 10 Nov 2026. - **Data tool:** [Secoda](/mcp/secoda/) - **Publisher:** Secoda - **Kind:** Official - **Repository or docs:** [github.com/secoda/secoda-mcp](https://github.com/secoda/secoda-mcp) - **Version or commit:** No releases; main branch last commit 26 Jan 2026 - **Last release:** none found on 10 Nov 2026 - **Transport:** Remote and local - **Auth:** Secoda API token (API_TOKEN env var) for the local FastMCP wrapper; docs also describe a Secoda-hosted MCP endpoint authenticated against workspace and AI permissions. - **Access:** Not stated in the source read - **Tools listed:** 8 named in the evidence - **Tool definitions read at:** [github.com/secoda/secoda-mcp/blob/main/server.py](https://github.com/secoda/secoda-mcp/blob/main/server.py) - **Context weight:** 63 tokens for 6 tool definitions (6 with a description), measured from the source code read - **Status:** Exists on 10 Nov 2026: open-source wrapper with 10 commits and no GitHub releases, plus a Secoda-hosted MCP described in docs.secoda.co (eight tools listed there; the wrapper code implements run_sql, search_data_assets, search_documentation, retrieve_entity, entity_lineage and glossary). - **Read:** 10 Nov 2026 ## Secoda MCP Secoda's server combines catalog search, glossary, lineage and SQL over a workspace, as a hosted endpoint and an open-source wrapper. `search_data_assets`, `search`, `run_sql`, `retrieve_entity`, `entity_lineage`, `glossary`, `get_secoda_docs`, `chart` - None found on 10 Nov 2026: Documentation of column-level lineage, transformation SQL, BI-field or metric lineage in the Secoda MCP docs. - None found on 10 Nov 2026: Ownership, classification or governance-rule propagation along lineage in the Secoda MCP docs. - None found on 10 Nov 2026: A list of supported warehouses or BI tools on the Secoda MCP page. - None found on 10 Nov 2026: Freshness statement in the Secoda MCP docs. ## Read next - [Secoda MCP, the full entry](/servers/secoda-mcp/): 17 facts - [Secoda MCP and Select Star MCP Server, side by side](/compare/secoda-mcp-vs-select-star-mcp/) - [Catalogs and metadata](/categories/catalogs/) ## Sources - docs.secoda.co/features/ai-assistant/secoda-mcp-server.md: https://docs.secoda.co/features/ai-assistant/secoda-mcp-server.md (read 10 Nov 2026) - github.com/secoda/secoda-mcp/releases: https://github.com/secoda/secoda-mcp/releases (read 10 Nov 2026) - api.github.com/repos/secoda/secoda-mcp/commits?per_page=1: https://api.github.com/repos/secoda/secoda-mcp/commits?per_page=1 (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/mcp/secoda/ --- # One MCP server for Select Star: Select Star MCP Server, from Select Star MCP SERVERS FOR A DATA TOOL, read 10 Nov 2026 **The short answer.** One MCP server for Select Star is listed: Select Star MCP Server, official, published by Select Star and read on 10 Nov 2026. It runs remote (hosted) and names 7 tools; the version read is Hosted, no version published. Last release: none found on 10 Nov 2026. - **Data tool:** [Select Star](/mcp/select-star/) - **Publisher:** Select Star (Snowflake agreed to acquire it, announced 24 Nov 2025) - **Kind:** Official - **Repository or docs:** [docs.selectstar.com/features/mcp-server](https://docs.selectstar.com/features/mcp-server) - **Version or commit:** Hosted, no version published - **Last release:** none found on 10 Nov 2026 - **Transport:** Remote (hosted) - **Auth:** Select Star API token sent as Authorization: Bearer header (mcp-remote proxy used for Claude Desktop). - **Access:** Read only - **Tools listed:** 7 named in the evidence - **Tool definitions read at:** [docs.selectstar.com/features/mcp-server](https://docs.selectstar.com/features/mcp-server) - **Context weight:** 98 tokens for 7 tool definitions (7 with a description), measured from the docs page read - **Status:** Exists on 10 Nov 2026 at mcp.production.selectstar.com/mcp. docs.selectstar.com did not resolve from the runner today, so the docs page was read from a web.archive.org snapshot dated 14 Jun 2026; Snowflake announced plans to acquire Select Star on 24 Nov 2025. - **Read:** 10 Nov 2026 ## Select Star MCP Server Select Star's hosted, read-only server exposes catalog search, lineage, query history and semantic models with token setup. `search_metadata`, `get_asset_details`, `get_table_info`, `traverse_lineage`, `get_asset_queries`, `search_semantic_models`, `get_semantic_model_details` - None found on 10 Nov 2026: Documentation on the MCP page that traverse_lineage is column-level or returns transformation SQL, BI-field or metric lineage. - None found on 10 Nov 2026: A statement that Select Star MCP results are limited to the asking user's own permissions. - None found on 10 Nov 2026: Any freshness or collection-cadence statement on the Select Star MCP docs page. ## Read next - [Select Star MCP Server, the full entry](/servers/select-star-mcp/): 16 facts - [Secoda MCP and Select Star MCP Server, side by side](/compare/secoda-mcp-vs-select-star-mcp/) - [Catalogs and metadata](/categories/catalogs/) ## Sources - selectstar.com/product/mcp-for-data: https://selectstar.com/product/mcp-for-data (read 10 Nov 2026) - docs.selectstar.com/features/mcp-server: https://docs.selectstar.com/features/mcp-server (read 10 Nov 2026) - web.archive.org/web/20260614045618/https://docs.selectstar.com/features/mcp-server: https://web.archive.org/web/20260614045618/https://docs.selectstar.com/features/mcp-server (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/mcp/select-star/ --- # One MCP server for Sifflet: Sifflet MCP Server, from Sifflet MCP SERVERS FOR A DATA TOOL, read 10 Nov 2026 **The short answer.** One MCP server for Sifflet is listed: Sifflet MCP Server, official, published by Sifflet and read on 10 Nov 2026. It runs local (stdio) and names 10 tools; the version read is 0.1.7 (PyPI); main at commit 1f390ca. Last release: 3 Aug 2026. 12 sourced facts follow, each with its read date. - **Data tool:** [Sifflet](/mcp/sifflet/) - **Publisher:** Sifflet - **Kind:** Official - **Repository or docs:** [github.com/siffletdata/sifflet-mcp](https://github.com/siffletdata/sifflet-mcp) - **Version or commit:** 0.1.7 (PyPI); main at commit 1f390ca - **Last release:** 3 Aug 2026 - **Transport:** Local (stdio) - **Auth:** Sifflet API token (Viewer role for most tools, Editor for monitor generation) and backend URL as environment variables. - **Access:** Read and write - **Tools listed:** 10 named in the evidence - **Tool definitions read at:** [raw.githubusercontent.com/siffletdata/sifflet-mcp/1f390ca06d7919214bfe067991539502fdb4b9fc/src/sifflet_mcp/server.py](https://raw.githubusercontent.com/siffletdata/sifflet-mcp/1f390ca06d7919214bfe067991539502fdb4b9fc/src/sifflet_mcp/server.py) - **Context weight:** 662 tokens for 10 tool definitions (10 with a description), measured from the source code read - **Status:** Exists on 10 Nov 2026; Sifflet docs page (updated 1 Oct 2026) points to the siffletdata/sifflet-mcp GitHub repository, MIT license, PyPI 0.1.7 released 3 Aug 2026; listed after the owner asked for one re-check (the docs host docs.sifflet.io did not load on 10 Nov 2026, docs.siffletdata.com did). - **Read:** 10 Nov 2026 ## Sifflet MCP Server Sifflet's local server covers asset search, incidents, monitors as code and downstream impact from one Sifflet API token. `asset_by_urn`, `search_asset`, `search_incidents`, `get_incident_by_issue_number`, `get_incident_scope_by_issue_number`, `get_monitor_details_by_id`, `close_incident_by_id`, `open_incident_by_id`, `get_monitor_code_by_description`, `get_downstream_assets_of_asset` - None found on 10 Nov 2026: An audit log or per-call permission prompt described for the MCP server in the README or the docs page. ## Read next - [Sifflet MCP Server, the full entry](/servers/sifflet-mcp/): 12 facts - [Monte Carlo MCP Server and Sifflet MCP Server, side by side](/compare/monte-carlo-mcp-vs-sifflet-mcp/) - [Data observability](/categories/observability/) ## Sources - github.com/siffletdata/sifflet-mcp/blob/main/README.md: https://github.com/siffletdata/sifflet-mcp/blob/main/README.md (read 10 Nov 2026) - raw.githubusercontent.com/siffletdata/sifflet-mcp/main/pyproject.toml: https://raw.githubusercontent.com/siffletdata/sifflet-mcp/main/pyproject.toml (read 10 Nov 2026) - pypi.org/project/sifflet-mcp: https://pypi.org/project/sifflet-mcp/ (read 10 Nov 2026) - docs.siffletdata.com/docs/sifflet-mcp-server.md: https://docs.siffletdata.com/docs/sifflet-mcp-server.md (read 10 Nov 2026) - github.com/siffletdata/sifflet-mcp/blob/1f390ca06d7919214bfe067991539502fdb4b9fc/src/sifflet_mcp/server.py: https://github.com/siffletdata/sifflet-mcp/blob/1f390ca06d7919214bfe067991539502fdb4b9fc/src/sifflet_mcp/server.py (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/mcp/sifflet/ --- # One MCP server for Sigma: Sigma MCP server, from Sigma Computing MCP SERVERS FOR A DATA TOOL, read 10 Nov 2026 **The short answer.** One MCP server for Sigma is listed: Sigma MCP server, official, published by Sigma Computing and read on 10 Nov 2026. It runs remote (hosted) and names 0 tools; the version read is Hosted service (no versioned release). Last release: none found on 10 Nov 2026. - **Data tool:** [Sigma](/mcp/sigma/) - **Publisher:** Sigma Computing - **Kind:** Official - **Repository or docs:** [help.sigmacomputing.com/docs/use-sigma-mcp-server](https://help.sigmacomputing.com/docs/use-sigma-mcp-server) - **Version or commit:** Hosted service (no versioned release) - **Last release:** none found on 10 Nov 2026 - **Transport:** Remote (hosted) - **Auth:** OAuth (HTTP transport) inheriting the user's Sigma account permissions; API-key access token with scope mcp:access (Beta). - **Access:** Read and write - **Tools listed:** 0 named in the evidence - **Tool definitions read at:** [help.sigmacomputing.com/docs/use-sigma-mcp-server.md](https://help.sigmacomputing.com/docs/use-sigma-mcp-server.md) - **Context weight:** Not measured: The docs page describes features (search, explore, analyze, build workbooks, screenshots, agents) in prose with example prompts but publishes no tool names or per-tool list. - **Status:** Exists on 10 Nov 2026 as a Sigma-hosted MCP server (URL per organization under Profile > Integrations). The docs describe capabilities (search, describe, query, build workbooks, screenshots, call agents) but publish no tool names, so tools is empty. - **Read:** 10 Nov 2026 ## Sigma MCP server Sigma's hosted server searches data models and workbooks, reads column definitions, and queries and builds workbooks under the user's own Sigma permissions. No tool names are published in the source read on 10 Nov 2026. - None found on 10 Nov 2026: Lineage or upstream-dependency capability on the Sigma MCP server page. - None found on 10 Nov 2026: Scheduled or event-driven metadata collection on the Sigma MCP page; queries run at call time. ## Read next - [Sigma MCP server, the full entry](/servers/sigma-mcp/): 18 facts - [Sigma MCP server and ThoughtSpot MCP Server, side by side](/compare/sigma-mcp-vs-thoughtspot-mcp-server/) - [BI and analytics](/categories/bi/) ## Sources - help.sigmacomputing.com/docs/use-sigma-mcp-server: https://help.sigmacomputing.com/docs/use-sigma-mcp-server (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/mcp/sigma/ --- # 3 MCP servers for Snowflake, read from their own sources MCP SERVERS FOR A DATA TOOL, read 10 Nov 2026 **The short answer.** 3 MCP servers for Snowflake are listed, each read from its own source on 10 Nov 2026: mcp-snowflake-server (community) (community, local (stdio)); Snowflake Labs MCP Server (official, labs org, remote and local); Snowflake-managed MCP server (official, remote (hosted)). Between them they name 20 tools. | Field | mcp-snowflake-server (community) | Snowflake Labs MCP Server | Snowflake-managed MCP server | | --- | --- | --- | --- | | Data tool | [Snowflake](/mcp/snowflake/) | [Snowflake](/mcp/snowflake/) | [Snowflake](/mcp/snowflake/) | | Publisher | isaacwasserman (community) | Snowflake Labs | Snowflake | | Kind | Community | Official, Labs org | Official | | Repository or docs | [github.com/isaacwasserman/mcp-snowflake-server](https://github.com/isaacwasserman/mcp-snowflake-server) | [github.com/Snowflake-Labs/mcp](https://github.com/Snowflake-Labs/mcp) | [docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-mcp](https://docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-mcp) | | Version or commit | 0.4.0 | 1.4.2 | Docs page, MCP revision 25 Nov 2025 | | Last release | 10 Apr 2025 | 15 May 2026 | none found on 10 Nov 2026 | | Transport | Local (stdio) | Remote and local | Remote (hosted) | | Auth | Snowflake user credentials (password, private key, external browser) or connections.toml. | Snowflake Python connector: password/PAT, key pair, OAuth, SSO, MFA; honors the connection role's RBAC. | Snowflake OAuth (default) or External OAuth; PAT supported; RBAC per MCP server and per tool. | | Access | Configurable | Configurable | Configurable | | Tools listed | 8 named in the evidence | 6 named in the evidence | 6 named in the evidence | | Tool definitions read at | [raw.githubusercontent.com/isaacwasserman/mcp-snowflake-server/main/README.md](https://raw.githubusercontent.com/isaacwasserman/mcp-snowflake-server/main/README.md) | [raw.githubusercontent.com/Snowflake-Labs/mcp/main/README.md](https://raw.githubusercontent.com/Snowflake-Labs/mcp/main/README.md) | [docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-mcp](https://docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-mcp) | | Context weight | 103 tokens for 8 tool definitions (8 with a description), measured from the README read | Not measured: The Snowflake Labs server, marked deprecated in its README, which describes config-driven tool groups (Cortex Search, Analyst, Agent, object management, SQL execution, semantic views) but publishes no per-tool names or descriptions; facts.json lists these six groups. | Not measured: The docs page defines user-configured tool types (CORTEX_AGENT_RUN, CODING_AGENT, CORTEX_SEARCH_SERVICE_QUERY, CORTEX_ANALYST_MESSAGE, SYSTEM_EXECUTE_SQL, GENERIC) used to build tools in a CREATE MCP SERVER specification; it publishes no fixed list of named tools, so facts.json lists tool types not tools. | | Status | Exists as a community server; PyPI 0.4.0 dated 10 Apr 2025, GitHub pushed 7 Oct 2025 (GitHub API). | Exists, but the README states it is deprecated and no longer maintained and points to the Snowflake-managed MCP server (read 10 Nov 2026). | Documented by Snowflake as the Snowflake-managed MCP server (read 10 Nov 2026); the Snowflake-Labs repository points users here. | | Read | 10 Nov 2026 | 10 Nov 2026 | 10 Nov 2026 | 3 servers for Snowflake, the same fields in the same order, read 10 Nov 2026. ## mcp-snowflake-server (community) A community server for schema browsing and read-only SQL on Snowflake, with write tools off by default. `read_query`, `write_query`, `create_table`, `list_databases`, `list_schemas`, `list_tables`, `describe_table`, `append_insight` - None found on 10 Nov 2026: A lineage or dependency tool in the README tool list. ## Snowflake Labs MCP Server The Labs server gives config-driven access to Cortex services, object management and semantic views under the connecting role; its README marks it deprecated. `Cortex Search services`, `Cortex Analyst services`, `Cortex Agent services`, `object manager group`, `query manager (SQL execution) group`, `semantic manager group` - None found on 10 Nov 2026: A lineage or column-lineage tool in the Snowflake-Labs/mcp README (checked tool and capability list). ## Snowflake-managed MCP server Snowflake's hosted endpoint applies RBAC per server and per tool with OAuth, and serves Cortex Analyst over semantic views, Cortex Search and SQL. `CORTEX_AGENT_RUN`, `CODING_AGENT`, `CORTEX_SEARCH_SERVICE_QUERY`, `CORTEX_ANALYST_MESSAGE`, `SYSTEM_EXECUTE_SQL`, `GENERIC` - None found on 10 Nov 2026: A lineage, column-lineage or object-dependency tool type on the Snowflake-managed MCP server page (tool types list read). ## Read next - [mcp-snowflake-server (community), the full entry](/servers/isaacwasserman-mcp-snowflake-server/): 13 facts - [Snowflake Labs MCP Server, the full entry](/servers/snowflake-labs-mcp/): 13 facts - [Snowflake-managed MCP server, the full entry](/servers/snowflake-managed-mcp/): 14 facts - [Snowflake Labs MCP Server and Snowflake-managed MCP server, side by side](/compare/snowflake-labs-mcp-vs-snowflake-managed-mcp/) - [mcp-snowflake-server (community) and Snowflake-managed MCP server, side by side](/compare/isaacwasserman-mcp-snowflake-server-vs-snowflake-managed-mcp/) - [mcp-snowflake-server (community) and Snowflake Labs MCP Server, side by side](/compare/isaacwasserman-mcp-snowflake-server-vs-snowflake-labs-mcp/) - [Warehouses](/categories/warehouses/) ## Sources - raw.githubusercontent.com/isaacwasserman/mcp-snowflake-server/main/README.md: https://raw.githubusercontent.com/isaacwasserman/mcp-snowflake-server/main/README.md (read 10 Nov 2026) - pypi.org/pypi/mcp-snowflake-server/json: https://pypi.org/pypi/mcp-snowflake-server/json (read 10 Nov 2026) - raw.githubusercontent.com/Snowflake-Labs/mcp/main/README.md: https://raw.githubusercontent.com/Snowflake-Labs/mcp/main/README.md (read 10 Nov 2026) - github.com/Snowflake-Labs/mcp/releases: https://github.com/Snowflake-Labs/mcp/releases (read 10 Nov 2026) - pypi.org/pypi/snowflake-labs-mcp/json: https://pypi.org/pypi/snowflake-labs-mcp/json (read 10 Nov 2026) - docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-mcp: https://docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-mcp (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/mcp/snowflake/ --- # One MCP server for SYNQ: SYNQ MCP (Coalesce Quality), from SYNQ MCP SERVERS FOR A DATA TOOL, read 10 Nov 2026 **The short answer.** One MCP server for SYNQ is listed: SYNQ MCP (Coalesce Quality), official, published by SYNQ and read on 10 Nov 2026. It runs remote (hosted) and names 134 tools; the version read is Hosted, no version published. Last release: none found on 10 Nov 2026. - **Data tool:** [SYNQ](/mcp/synq/) - **Publisher:** SYNQ (docs titled Coalesce Quality) - **Kind:** Official - **Repository or docs:** [docs.synq.io/scout/mcp.md](https://docs.synq.io/scout/mcp.md) - **Version or commit:** Hosted, no version published - **Last release:** none found on 10 Nov 2026 - **Transport:** Remote (hosted) - **Auth:** OAuth 2 with PKCE per user with selectable permission tier (read, write, governance, deploy); legacy workspace token (st-...) as Bearer. - **Access:** Configurable - **Tools listed:** 134 named in the evidence - **Tool definitions read at:** [docs.synq.io/scout/mcp.md](https://docs.synq.io/scout/mcp.md) - **Context weight:** 4,550 tokens for 134 tool definitions (134 with a description), measured from the docs page read - **Status:** Exists on 10 Nov 2026; docs.synq.io now titles itself Coalesce Quality Documentation and its MCP page calls the product Coalesce Quality; regional hosted endpoints EU/US/AU. - **Read:** 10 Nov 2026 ## SYNQ MCP (Coalesce Quality) SYNQ's hosted observability and lineage server uses per-user OAuth with tiered consent scopes, and tools hide when a scope is not granted. `add_comment`, `apply_service_account_changes`, `batch_entity_annotations`, `batch_entity_details`, `batch_executions`, `batch_urls`, `browse_folders`, `create_incident`, `create_monitors`, `create_sql_tests`, `create_table_stats_monitor`, `decide_recommendation`, `delete_alert`, `delete_check_categorisation_rule`, `delete_data_product`, `delete_deployment_rule`, `delete_domain`, `delete_monitors`, `delete_owner`, `delete_saved_view`, `delete_sql_test_deployment_rule`, `delete_sql_tests`, `deploy_deployment_rule`, `deploy_sql_test_deployment_rule`, `describe_check_categorisation`, `describe_check_types`, `describe_resolver_ql`, `exclude_from_monitoring`, `execute_monitor`, `folder_of`, `get_alert`, `get_code`, `get_commit_diff`, `get_common_upstream_dependencies`, `get_coverage_summary`, `get_data_product`, `get_database_change_details`, `get_database_coordinates_from_entity`, `get_deployment_rule`, `get_domain`, `get_downstream_dependencies`, `get_entity_access`, `get_entity_details`, `get_entity_from_database_coordinates`, `get_entity_impact`, `get_entity_knowledge`, `get_entity_metrics`, `get_entity_overlays`, `get_immediate_downstream_dependencies`, `get_immediate_upstream_dependencies`, `get_incident_details`, `get_issue`, `get_issue_comments`, `get_issue_details`, `get_issue_impact`, `get_join_patterns`, `get_knowledge_pages`, `get_latest_executions`, `get_lineage`, `get_monitor`, `get_monitor_predictions`, `get_owner`, `get_query_activity`, `get_saved_view`, `get_schema`, `get_sql_test_deployment_rule`, `get_upstream_dependencies`, `get_upstream_sources`, `iam`, `list_access_signals`, `list_alerts`, `list_annotations`, `list_categorised_checks`, `list_check_categories`, `list_check_categorisation_rules`, `list_checks`, `list_commits`, `list_data_product_members`, `list_data_products`, `list_database_changes`, `list_deployment_rules`, `list_domain_members`, `list_domains`, `list_entity_coverage`, `list_executions`, `list_folder_members`, `list_history`, `list_integration_health`, `list_knowledge_pages`, `list_monitor_segments`, `list_monitoring_exclusions`, `list_monitors`, `list_open_incidents`, `list_open_issues`, `list_owners`, `list_ownerships`, `list_platform_users`, `list_recommendations`, `list_saved_views`, `list_service_account_suggestions`, `list_service_accounts`, `list_sql_test_deployment_rules`, `list_sql_tests`, `list_table_stats_monitors`, `list_users`, `plan_service_account_changes`, `preview_check_categorisation_rule`, `preview_deployment_rule`, `preview_sql_test_deployment_rule`, `profile_columns`, `resolve_query`, `run_sql_test_again`, `sample_column_values`, `sample_entities`, `save_alert`, `save_check_categorisation_rule`, `save_data_product`, `save_domain`, `save_owner`, `save_saved_view`, `search_entities`, `send_test_alert`, `set_alert_enabled`, `set_check_category`, `set_issue_status`, `set_monitor_segment_labels`, `set_saved_view_pin`, `submit_feature_request`, `summarise_executions`, `unexclude_from_monitoring`, `update_incident`, `update_monitors`, `update_sql_tests`, `update_table_stats_monitor` - None found on 10 Nov 2026: A GitHub repository, package version or release date for the SYNQ MCP server on the docs page. ## Read next - [SYNQ MCP (Coalesce Quality), the full entry](/servers/synq-mcp/): 18 facts - [Monte Carlo MCP Server and SYNQ MCP (Coalesce Quality), side by side](/compare/monte-carlo-mcp-vs-synq-mcp/) - [Data observability](/categories/observability/) ## Sources - docs.synq.io/scout/mcp.md: https://docs.synq.io/scout/mcp.md (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/mcp/synq/ --- # One MCP server for Tableau: Tableau MCP, from Tableau MCP SERVERS FOR A DATA TOOL, read 10 Nov 2026 **The short answer.** One MCP server for Tableau is listed: Tableau MCP, official, published by Tableau and read on 10 Nov 2026. It runs remote and local and names 40 tools; the version read is 6.0.4. Last release: 9 Nov 2026. 25 sourced facts follow, each with its read date. - **Data tool:** [Tableau](/mcp/tableau/) - **Publisher:** Tableau (Salesforce) - **Kind:** Official - **Repository or docs:** [github.com/tableau/tableau-mcp](https://github.com/tableau/tableau-mcp) - **Version or commit:** 6.0.4 - **Last release:** 9 Nov 2026 - **Transport:** Remote and local - **Auth:** Hosted: OAuth 2.1 with each user's own Tableau Cloud identity. Self-hosted: PAT (default), direct-trust (Connected Apps), OAuth, passthrough, or unified access tokens. - **Access:** Configurable - **Tools listed:** 40 named in the evidence - **Tool definitions read at:** [raw.githubusercontent.com/tableau/tableau-mcp/main/src/tools/web/getDatasourceMetadata/getDatasourceMetadata.ts](https://raw.githubusercontent.com/tableau/tableau-mcp/main/src/tools/web/getDatasourceMetadata/getDatasourceMetadata.ts) - **Context weight:** Not measured: the source read on 10 Nov 2026 holds the definitions of 2 of the roughly 40 Tableau MCP tools (get-datasource-metadata and query-datasource), and one of those descriptions is imported from a file that was not read, so a measurement would cover a small part of the server. - **Status:** Exists on 10 Nov 2026 as tableau/tableau-mcp (npm @tableau/mcp-server) and as a hosted service at mcp.tableau.com; the schema file holds two of the tool source files (get-datasource-metadata, query-datasource); the full tool list comes from the docs intro page. - **Read:** 10 Nov 2026 ## Tableau MCP Tableau's server exposes data source metadata with calculation strings and Pulse metric definitions per signed-in user, plus, on Tableau+, a knowledge tool for lineage and impact. `list-datasources`, `list-workbooks`, `list-projects`, `list-views`, `list-custom-views`, `get-datasource-metadata`, `get-workbook`, `download-workbook`, `get-flow`, `describe-flow`, `get-flow-task`, `list-flows`, `list-flow-runs`, `list-flow-tasks`, `run-flow`, `run-flow-task`, `cancel-flow-run`, `delete-content`, `get-view-data`, `get-view-image`, `get-custom-view-data`, `get-custom-view-image`, `query-datasource`, `list-all-pulse-metric-definitions`, `list-pulse-metric-definitions-from-definition-ids`, `list-pulse-metrics-from-metric-definition-id`, `list-pulse-metrics-from-metric-ids`, `list-pulse-metric-subscriptions`, `generate-pulse-metric-value-insight-bundle`, `generate-pulse-insight-brief`, `search-content`, `list-extract-refresh-tasks`, `update-cloud-extract-refresh-task`, `list-users`, `update-user`, `query-admin-insights`, `query-knowledge-context`, `inspect-knowledge-context`, `manage-knowledge-context`, `scaffold-data-app` - None found on 10 Nov 2026: Any scheduled or event-driven metadata collection in the Tableau MCP docs intro page; the server calls Tableau REST, Metadata and VizQL APIs per request. ## Read next - [Tableau MCP, the full entry](/servers/tableau-mcp/): 25 facts - [Power BI Modeling MCP Server and Tableau MCP, side by side](/compare/powerbi-modeling-mcp-vs-tableau-mcp/) - [Looker tools in MCP Toolbox for Databases and Tableau MCP, side by side](/compare/looker-mcp-toolbox-vs-tableau-mcp/) - [BI and analytics](/categories/bi/) ## Sources - tableau.github.io/tableau-mcp/docs/intro: https://tableau.github.io/tableau-mcp/docs/intro (read 10 Nov 2026) - github.com/tableau/tableau-mcp/releases: https://github.com/tableau/tableau-mcp/releases (read 10 Nov 2026) - registry.npmjs.org/@tableau/mcp-server: https://registry.npmjs.org/@tableau/mcp-server (read 10 Nov 2026) - raw.githubusercontent.com/tableau/tableau-mcp/main/README.md: https://raw.githubusercontent.com/tableau/tableau-mcp/main/README.md (read 10 Nov 2026) - github.com/tableau/tableau-mcp: https://github.com/tableau/tableau-mcp (read 10 Nov 2026) - tableau.github.io/tableau-mcp/docs/configuration/mcp-config/env-vars: https://tableau.github.io/tableau-mcp/docs/configuration/mcp-config/env-vars (read 10 Nov 2026) - tableau.github.io/tableau-mcp/docs/configuration/mcp-config/authentication: https://tableau.github.io/tableau-mcp/docs/configuration/mcp-config/authentication (read 10 Nov 2026) - raw.githubusercontent.com/tableau/tableau-mcp/main/src/tools/web/queryDatasource/queryDatasource.ts: https://raw.githubusercontent.com/tableau/tableau-mcp/main/src/tools/web/queryDatasource/queryDatasource.ts (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/mcp/tableau/ --- # One MCP server for ThoughtSpot: ThoughtSpot MCP Server, from ThoughtSpot MCP SERVERS FOR A DATA TOOL, read 10 Nov 2026 **The short answer.** One MCP server for ThoughtSpot is listed: ThoughtSpot MCP Server, official, published by ThoughtSpot and read on 10 Nov 2026. It runs remote (hosted) and names 20 tools; the version read is 0.5.0 (npm and package.json); main at commit 820bf16f6aa3. Last release: 6 Jun 2025. - **Data tool:** [ThoughtSpot](/mcp/thoughtspot/) - **Publisher:** ThoughtSpot - **Kind:** Official - **Repository or docs:** [github.com/thoughtspot/mcp-server](https://github.com/thoughtspot/mcp-server) - **Version or commit:** 0.5.0 (npm and package.json); main at commit 820bf16f6aa3 - **Last release:** 6 Jun 2025 - **Transport:** Remote (hosted) - **Auth:** OAuth to a ThoughtSpot instance (dynamic client registration, or manually registered client id and secret); bearer-token and /sse paths also exist; the user chooses or is bound to a ThoughtSpot instance URL. - **Access:** Read and write - **Tools listed:** 20 named in the evidence - **Tool definitions read at:** [raw.githubusercontent.com/thoughtspot/mcp-server/main/src/servers/tool-definitions.ts](https://raw.githubusercontent.com/thoughtspot/mcp-server/main/src/servers/tool-definitions.ts) - **Context weight:** at least 2,149 tokens for 20 tool definitions (20 with a description), measured from the source code read - **Status:** Exists on 10 Nov 2026 as thoughtspot/mcp-server, run as a Cloudflare Worker (hosted at agent.thoughtspot.app) with a local stdio entry in package.json. The GitHub repo has no releases or tags; the npm package last published 6 Jun 2025 although main has moved on, so last_release comes from npm. - **Read:** 10 Nov 2026 ## ThoughtSpot MCP Server ThoughtSpot's server returns Spotter answers, searches objects with owner, tag and verified filters, and creates dashboards and models within the user's privileges. `ping`, `getRelevantQuestions`, `getAnswer`, `createLiveboard`, `getDataSourceSuggestions`, `check_connectivity`, `search_objects`, `get_data`, `create_analysis_session`, `send_session_message`, `get_session_updates`, `create_dashboard`, `modify_dashboard`, `get_dashboard_status`, `list_orgs`, `switch_org`, `create_model_session`, `send_model_message`, `get_model_updates`, `finalize_model` - None found on 10 Nov 2026: A lineage tool in tool-definitions.ts (no column, upstream-table or transformation lineage tool among the 20 tools). - None found on 10 Nov 2026: Scheduled or event-driven metadata collection in the ThoughtSpot MCP README; tools query the instance per request. ## Read next - [ThoughtSpot MCP Server, the full entry](/servers/thoughtspot-mcp-server/): 19 facts - [Sigma MCP server and ThoughtSpot MCP Server, side by side](/compare/sigma-mcp-vs-thoughtspot-mcp-server/) - [BI and analytics](/categories/bi/) ## Sources - github.com/thoughtspot/mcp-server: https://github.com/thoughtspot/mcp-server (read 10 Nov 2026) - github.com/thoughtspot/mcp-server/releases: https://github.com/thoughtspot/mcp-server/releases (read 10 Nov 2026) - registry.npmjs.org/@thoughtspot/mcp-server: https://registry.npmjs.org/@thoughtspot/mcp-server (read 10 Nov 2026) - raw.githubusercontent.com/thoughtspot/mcp-server/main/README.md: https://raw.githubusercontent.com/thoughtspot/mcp-server/main/README.md (read 10 Nov 2026) - raw.githubusercontent.com/thoughtspot/mcp-server/main/src/servers/tool-definitions.ts: https://raw.githubusercontent.com/thoughtspot/mcp-server/main/src/servers/tool-definitions.ts (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/mcp/thoughtspot/ --- # 26 data tools, 30 MCP servers, filed by tool name DATA TOOLS, read 10 Nov 2026 **The short answer.** The directory files 30 MCP servers under 26 data tools, alphabetically from Airbyte to ThoughtSpot. Snowflake has 3 servers listed, BigQuery and Databricks 2 each, and every other tool one. Each page puts the same fourteen fields side by side, read 10 Nov 2026. | Data tool | Kind of tool | Servers | Listed | | --- | --- | --- | --- | | [Airbyte](/mcp/airbyte/) | Ingestion connectors | 1 | [Airbyte MCP (hosted)](/servers/airbyte-mcp/) | | [Alation](/mcp/alation/) | Catalogs and metadata | 1 | [Alation AI Agent SDK MCP server](/servers/alation-mcp/) | | [Amazon Redshift](/mcp/redshift/) | Warehouses | 1 | [Amazon Redshift MCP Server (AWS Labs)](/servers/awslabs-redshift-mcp-server/) | | [Atlan](/mcp/atlan/) | Catalogs and metadata | 1 | [Atlan MCP Server](/servers/atlan-mcp/) | | [BigQuery](/mcp/bigquery/) | Warehouses | 2 | [BigQuery remote MCP server](/servers/bigquery-remote-mcp/); [MCP Toolbox for Databases (BigQuery tools)](/servers/mcp-toolbox-bigquery/) | | [Collibra](/mcp/collibra/) | Catalogs and metadata | 1 | [Collibra MCP Server (chip)](/servers/collibra-mcp/) | | [Cube](/mcp/cube/) | BI and analytics | 1 | [Cube MCP server](/servers/cube-mcp/) | | [Databricks](/mcp/databricks/) | Warehouses | 2 | [Databricks Labs MCP servers](/servers/databrickslabs-mcp/); [Databricks managed MCP servers](/servers/databricks-managed-mcp/) | | [DataHub](/mcp/datahub/) | Catalogs and metadata | 1 | [DataHub MCP Server](/servers/datahub-mcp/) | | [dbt](/mcp/dbt/) | Transformation | 1 | [dbt MCP Server](/servers/dbt-mcp/) | | [Euno](/mcp/euno/) | Catalogs and metadata | 1 | [Euno MCP](/servers/euno-mcp/) | | [Fivetran](/mcp/fivetran/) | Ingestion connectors | 1 | [Fivetran MCP Server](/servers/fivetran-mcp/) | | [Hex](/mcp/hex/) | BI and analytics | 1 | [Hex MCP server](/servers/hex-mcp/) | | [Looker](/mcp/looker/) | BI and analytics | 1 | [Looker tools in MCP Toolbox for Databases](/servers/looker-mcp-toolbox/) | | [Monte Carlo](/mcp/monte-carlo/) | Data observability | 1 | [Monte Carlo MCP Server](/servers/monte-carlo-mcp/) | | [Omni](/mcp/omni/) | BI and analytics | 1 | [Omni MCP Server](/servers/omni-mcp/) | | [OpenMetadata](/mcp/openmetadata/) | Catalogs and metadata | 1 | [OpenMetadata MCP Server](/servers/openmetadata-mcp/) | | [Power BI](/mcp/power-bi/) | BI and analytics | 1 | [Power BI Modeling MCP Server](/servers/powerbi-modeling-mcp/) | | [Secoda](/mcp/secoda/) | Catalogs and metadata | 1 | [Secoda MCP](/servers/secoda-mcp/) | | [Select Star](/mcp/select-star/) | Catalogs and metadata | 1 | [Select Star MCP Server](/servers/select-star-mcp/) | | [Sifflet](/mcp/sifflet/) | Data observability | 1 | [Sifflet MCP Server](/servers/sifflet-mcp/) | | [Sigma](/mcp/sigma/) | BI and analytics | 1 | [Sigma MCP server](/servers/sigma-mcp/) | | [Snowflake](/mcp/snowflake/) | Warehouses | 3 | [mcp-snowflake-server (community)](/servers/isaacwasserman-mcp-snowflake-server/); [Snowflake Labs MCP Server](/servers/snowflake-labs-mcp/); [Snowflake-managed MCP server](/servers/snowflake-managed-mcp/) | | [SYNQ](/mcp/synq/) | Data observability | 1 | [SYNQ MCP (Coalesce Quality)](/servers/synq-mcp/) | | [Tableau](/mcp/tableau/) | BI and analytics | 1 | [Tableau MCP](/servers/tableau-mcp/) | | [ThoughtSpot](/mcp/thoughtspot/) | BI and analytics | 1 | [ThoughtSpot MCP Server](/servers/thoughtspot-mcp-server/) | Filed alphabetically by data tool, read 10 Nov 2026. Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/mcp/ --- # 8 MCP servers for warehouses, filed by data tool CATEGORY, read 10 Nov 2026 **The short answer.** For warehouses, the directory lists 8 MCP servers across 4 data tools: Amazon Redshift, BigQuery, Databricks, Snowflake. 6 can run hosted and 5 locally; together they name 53 tools, read 10 Nov 2026. Every entry shows the same fourteen fields in the same order, from publisher and version to auth, access and context weight. Warehouse servers run SQL and read schemas for an agent. Snowflake has three servers listed (the Snowflake-managed server, the deprecated Snowflake Labs server and a community server); BigQuery and Databricks have two each; Amazon Redshift has one, from AWS Labs. | Data tool | Server | Kind | Transport | Access | Tools | Version | Last release | | --- | --- | --- | --- | --- | --- | --- | --- | | Amazon Redshift | [Amazon Redshift MCP Server (AWS Labs)](/servers/awslabs-redshift-mcp-server/) | Official, Labs org | Local (stdio) | Configurable | 8 | 0.1.3 | 9 Nov 2026 | | BigQuery | [BigQuery remote MCP server](/servers/bigquery-remote-mcp/) | Official | Remote (hosted) | Configurable | 9 | Docs reference last updated 3 Nov 2026 | none found | | BigQuery | [MCP Toolbox for Databases (BigQuery tools)](/servers/mcp-toolbox-bigquery/) | Official | Remote and local | Configurable | 9 | V1.16.0 | 5 Nov 2026 | | Databricks | [Databricks Labs MCP servers](/servers/databrickslabs-mcp/) | Official, Labs org | Remote and local | Not stated in the source read | 3 | none found on 10 Nov 2026 | none found | | Databricks | [Databricks managed MCP servers](/servers/databricks-managed-mcp/) | Official | Remote (hosted) | Configurable | 4 | Docs page last updated Nov 9, 2026 | none found | | Snowflake | [mcp-snowflake-server (community)](/servers/isaacwasserman-mcp-snowflake-server/) | Community | Local (stdio) | Configurable | 8 | 0.4.0 | 10 Apr 2025 | | Snowflake | [Snowflake Labs MCP Server](/servers/snowflake-labs-mcp/) | Official, Labs org | Remote and local | Configurable | 6 | 1.4.2 | 15 May 2026 | | Snowflake | [Snowflake-managed MCP server](/servers/snowflake-managed-mcp/) | Official | Remote (hosted) | Configurable | 6 | Docs page, MCP revision 25 Nov 2025 | none found | Warehouses: 8 servers, filed by data tool, read 10 Nov 2026. ## In one sentence each - [Amazon Redshift MCP Server (AWS Labs)](/servers/awslabs-redshift-mcp-server/): The AWS Labs server is read-only by default with an explicit read-write opt-in, confirmation prompts and metadata discovery over the Redshift Data API. - [BigQuery remote MCP server](/servers/bigquery-remote-mcp/): Google Cloud's managed endpoint gives IAM-scoped access to a SELECT-only query tool and dataset and table metadata tools. - [MCP Toolbox for Databases (BigQuery tools)](/servers/mcp-toolbox-bigquery/): Google's open-source Toolbox offers BigQuery SQL and metadata tools with a read-only mode, dataset allowlists and Knowledge Catalog search. - [Databricks Labs MCP servers](/servers/databrickslabs-mcp/): The Labs server exposes Unity Catalog functions, vector search indexes and Genie spaces as tools; its README marks it deprecated in favor of the managed servers. - [Databricks managed MCP servers](/servers/databricks-managed-mcp/): Databricks-provided MCP endpoints run under the caller's Unity Catalog grants and include Genie spaces for questions over a governed semantic layer. - [mcp-snowflake-server (community)](/servers/isaacwasserman-mcp-snowflake-server/): A community server for schema browsing and read-only SQL on Snowflake, with write tools off by default. - [Snowflake Labs MCP Server](/servers/snowflake-labs-mcp/): The Labs server gives config-driven access to Cortex services, object management and semantic views under the connecting role; its README marks it deprecated. - [Snowflake-managed MCP server](/servers/snowflake-managed-mcp/): Snowflake's hosted endpoint applies RBAC per server and per tool with OAuth, and serves Cortex Analyst over semantic views, Cortex Search and SQL. ## Other kinds of data tool - [Transformation](/categories/transformation/) - [Ingestion connectors](/categories/ingestion/) - [BI and analytics](/categories/bi/) - [Catalogs and metadata](/categories/catalogs/) - [Data observability](/categories/observability/) ## Sources - pypi.org/pypi/awslabs.redshift-mcp-server/json: https://pypi.org/pypi/awslabs.redshift-mcp-server/json (read 10 Nov 2026) - raw.githubusercontent.com/awslabs/mcp/main/src/redshift-mcp-server/README.md: https://raw.githubusercontent.com/awslabs/mcp/main/src/redshift-mcp-server/README.md (read 10 Nov 2026) - docs.cloud.google.com/bigquery/docs/use-bigquery-mcp: https://docs.cloud.google.com/bigquery/docs/use-bigquery-mcp (read 10 Nov 2026) - docs.cloud.google.com/bigquery/docs/reference/mcp: https://docs.cloud.google.com/bigquery/docs/reference/mcp (read 10 Nov 2026) - raw.githubusercontent.com/googleapis/mcp-toolbox/main/README.md: https://raw.githubusercontent.com/googleapis/mcp-toolbox/main/README.md (read 10 Nov 2026) - github.com/googleapis/mcp-toolbox/releases: https://github.com/googleapis/mcp-toolbox/releases (read 10 Nov 2026) - raw.githubusercontent.com/databrickslabs/mcp/master/README.md: https://raw.githubusercontent.com/databrickslabs/mcp/master/README.md (read 10 Nov 2026) - docs.databricks.com/aws/en/generative-ai/mcp/managed-mcp: https://docs.databricks.com/aws/en/generative-ai/mcp/managed-mcp (read 10 Nov 2026) - raw.githubusercontent.com/isaacwasserman/mcp-snowflake-server/main/README.md: https://raw.githubusercontent.com/isaacwasserman/mcp-snowflake-server/main/README.md (read 10 Nov 2026) - pypi.org/pypi/mcp-snowflake-server/json: https://pypi.org/pypi/mcp-snowflake-server/json (read 10 Nov 2026) - raw.githubusercontent.com/Snowflake-Labs/mcp/main/README.md: https://raw.githubusercontent.com/Snowflake-Labs/mcp/main/README.md (read 10 Nov 2026) - github.com/Snowflake-Labs/mcp/releases: https://github.com/Snowflake-Labs/mcp/releases (read 10 Nov 2026) - pypi.org/pypi/snowflake-labs-mcp/json: https://pypi.org/pypi/snowflake-labs-mcp/json (read 10 Nov 2026) - docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-mcp: https://docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-mcp (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/categories/warehouses/ --- # 1 MCP server for transformation, filed by data tool CATEGORY, read 10 Nov 2026 **The short answer.** For transformation, the directory lists 1 MCP server across 1 data tool: dbt. 1 can run hosted and 1 locally; together they name 52 tools, read 10 Nov 2026. Every entry shows the same fourteen fields in the same order, from publisher and version to auth, access and context weight. The dbt MCP Server from dbt Labs is the one transformation server listed. It combines the dbt CLI, the Discovery API, the Semantic Layer and SQL tools, locally or hosted. | Data tool | Server | Kind | Transport | Access | Tools | Version | Last release | | --- | --- | --- | --- | --- | --- | --- | --- | | dbt | [dbt MCP Server](/servers/dbt-mcp/) | Official | Remote and local | Configurable | 52 | 2.10.1 | 3 Nov 2026 | Transformation: 1 servers, filed by data tool, read 10 Nov 2026. ## In one sentence each - [dbt MCP Server](/servers/dbt-mcp/): dbt Labs' server combines the dbt CLI, the Discovery API with model and column lineage, the Semantic Layer and SQL tools on dbt Platform. ## Other kinds of data tool - [Warehouses](/categories/warehouses/) - [Ingestion connectors](/categories/ingestion/) - [BI and analytics](/categories/bi/) - [Catalogs and metadata](/categories/catalogs/) - [Data observability](/categories/observability/) ## Sources - github.com/dbt-labs/dbt-mcp/releases: https://github.com/dbt-labs/dbt-mcp/releases (read 10 Nov 2026) - docs.getdbt.com/docs/dbt-ai/about-mcp: https://docs.getdbt.com/docs/dbt-ai/about-mcp (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/categories/transformation/ --- # 2 MCP servers for ingestion connectors, filed by data tool CATEGORY, read 10 Nov 2026 **The short answer.** For ingestion connectors, the directory lists 2 MCP servers across 2 data tools: Airbyte, Fivetran. 2 can run hosted and 1 locally; together they name 103 tools, read 10 Nov 2026. Every entry shows the same fourteen fields in the same order, from publisher and version to auth, access and context weight. Ingestion connector servers let an agent inspect and operate pipelines: connections, schemas, sync status and runs. Airbyte and Fivetran each publish one. | Data tool | Server | Kind | Transport | Access | Tools | Version | Last release | | --- | --- | --- | --- | --- | --- | --- | --- | | Airbyte | [Airbyte MCP (hosted)](/servers/airbyte-mcp/) | Official | Remote (hosted) | Configurable | 50 | Hosted service (no versioned release); PyAirbyte package 0.71.3 ships an experimental local airbyte.mcp server | none found | | Fivetran | [Fivetran MCP Server](/servers/fivetran-mcp/) | Official | Remote and local | Configurable | 53 | V0.5.1 | 6 Oct 2026 | Ingestion connectors: 2 servers, filed by data tool, read 10 Nov 2026. ## In one sentence each - [Airbyte MCP (hosted)](/servers/airbyte-mcp/): Airbyte's hosted server lets an agent build, run and troubleshoot pipelines as the signed-in user, with a documented read-only mode; it serves pipeline and catalog state. - [Fivetran MCP Server](/servers/fivetran-mcp/): Fivetran's server mirrors its REST API behind a read-by-default scope with a deny-list, so an agent can inspect connections, schemas and sync status. ## Other kinds of data tool - [Warehouses](/categories/warehouses/) - [Transformation](/categories/transformation/) - [BI and analytics](/categories/bi/) - [Catalogs and metadata](/categories/catalogs/) - [Data observability](/categories/observability/) ## Sources - docs.airbyte.com/platform/airbyte-mcp.md: https://docs.airbyte.com/platform/airbyte-mcp.md (read 10 Nov 2026) - docs.airbyte.com/platform/airbyte-mcp/install.md: https://docs.airbyte.com/platform/airbyte-mcp/install.md (read 10 Nov 2026) - pypi.org/pypi/airbyte/json: https://pypi.org/pypi/airbyte/json (read 10 Nov 2026) - docs.airbyte.com/llms.txt: https://docs.airbyte.com/llms.txt (read 10 Nov 2026) - raw.githubusercontent.com/fivetran/fivetran-mcp/main/README.md: https://raw.githubusercontent.com/fivetran/fivetran-mcp/main/README.md (read 10 Nov 2026) - github.com/fivetran/fivetran-mcp/releases: https://github.com/fivetran/fivetran-mcp/releases (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/categories/ingestion/ --- # 8 MCP servers for bi and analytics, filed by data tool CATEGORY, read 10 Nov 2026 **The short answer.** For bi and analytics, the directory lists 8 MCP servers across 8 data tools: Cube, Hex, Looker, Omni, Power BI, Sigma, Tableau, ThoughtSpot. 8 can run hosted and 3 locally; together they name 210 tools, read 10 Nov 2026. Every entry shows the same fourteen fields in the same order, from publisher and version to auth, access and context weight. BI and analytics servers expose dashboards, workbooks, semantic models and natural-language answers. Eight are listed, from Cube and Hex to Tableau and ThoughtSpot; several can create or edit content under the user's own permissions. | Data tool | Server | Kind | Transport | Access | Tools | Version | Last release | | --- | --- | --- | --- | --- | --- | --- | --- | | Cube | [Cube MCP server](/servers/cube-mcp/) | Official | Remote (hosted) | Read and write | 30 | Hosted service (no versioned release) | none found | | Hex | [Hex MCP server](/servers/hex-mcp/) | Official | Remote (hosted) | Read and write | 18 | Hosted service (no versioned release) | none found | | Looker | [Looker tools in MCP Toolbox for Databases](/servers/looker-mcp-toolbox/) | Official | Remote and local | Configurable | 55 | V1.16.0 | 5 Nov 2026 | | Omni | [Omni MCP Server](/servers/omni-mcp/) | Official | Remote (hosted) | Configurable | 26 | Hosted service (no versioned release) | none found | | Power BI | [Power BI Modeling MCP Server](/servers/powerbi-modeling-mcp/) | Official | Remote and local | Configurable | 21 | 1.0.1 | 6 Nov 2026 | | Sigma | [Sigma MCP server](/servers/sigma-mcp/) | Official | Remote (hosted) | Read and write | 0 | Hosted service (no versioned release) | none found | | Tableau | [Tableau MCP](/servers/tableau-mcp/) | Official | Remote and local | Configurable | 40 | 6.0.4 | 9 Nov 2026 | | ThoughtSpot | [ThoughtSpot MCP Server](/servers/thoughtspot-mcp-server/) | Official | Remote (hosted) | Read and write | 20 | 0.5.0 (npm and package.json); main at commit 820bf16f6aa3 | 6 Jun 2025 | BI and analytics: 8 servers, filed by data tool, read 10 Nov 2026. ## In one sentence each - [Cube MCP server](/servers/cube-mcp/): Cube's hosted server searches the semantic layer, runs SQL with row-level security as the signed-in user and edits the semantic model on a branch. - [Hex MCP server](/servers/hex-mcp/): Hex's hosted server runs Threads (agent analyses over the user's data connections) and notebook editing under workspace roles and project permissions. - [Looker tools in MCP Toolbox for Databases](/servers/looker-mcp-toolbox/): Google's Toolbox exposes Looker models, explores, dimensions, measures and LookML project files, plus instance health and usage tools. - [Omni MCP Server](/servers/omni-mcp/): Omni's hosted server runs as a specific Omni user and exposes models, topics, saved queries and apps, so answers follow the semantic model and the user's permissions. - [Power BI Modeling MCP Server](/servers/powerbi-modeling-mcp/): Microsoft's authoring server reads and writes semantic-model objects (tables, columns, DAX measures, relationships, RLS roles, TMDL) under the user's Fabric permissions, with a read-only switch. - [Sigma MCP server](/servers/sigma-mcp/): Sigma's hosted server searches data models and workbooks, reads column definitions, and queries and builds workbooks under the user's own Sigma permissions. - [Tableau MCP](/servers/tableau-mcp/): Tableau's server exposes data source metadata with calculation strings and Pulse metric definitions per signed-in user, plus, on Tableau+, a knowledge tool for lineage and impact. - [ThoughtSpot MCP Server](/servers/thoughtspot-mcp-server/): ThoughtSpot's server returns Spotter answers, searches objects with owner, tag and verified filters, and creates dashboards and models within the user's privileges. ## Other kinds of data tool - [Warehouses](/categories/warehouses/) - [Transformation](/categories/transformation/) - [Ingestion connectors](/categories/ingestion/) - [Catalogs and metadata](/categories/catalogs/) - [Data observability](/categories/observability/) ## Sources - docs.cube.dev/docs/integrations/mcp-server.md: https://docs.cube.dev/docs/integrations/mcp-server.md (read 10 Nov 2026) - learn.hex.tech/docs/api-integrations/mcp-server: https://learn.hex.tech/docs/api-integrations/mcp-server (read 10 Nov 2026) - raw.githubusercontent.com/googleapis/genai-toolbox/main/README.md: https://raw.githubusercontent.com/googleapis/genai-toolbox/main/README.md (read 10 Nov 2026) - github.com/googleapis/mcp-toolbox/releases: https://github.com/googleapis/mcp-toolbox/releases (read 10 Nov 2026) - docs.omni.co/ai/mcp/tools.md: https://docs.omni.co/ai/mcp/tools.md (read 10 Nov 2026) - learn.microsoft.com/en-us/power-bi/developer/mcp/power-bi-authoring-mcp: https://learn.microsoft.com/en-us/power-bi/developer/mcp/power-bi-authoring-mcp (read 10 Nov 2026) - github.com/microsoft/powerbi-modeling-mcp/releases: https://github.com/microsoft/powerbi-modeling-mcp/releases (read 10 Nov 2026) - registry.npmjs.org/@microsoft/powerbi-modeling-mcp: https://registry.npmjs.org/@microsoft/powerbi-modeling-mcp (read 10 Nov 2026) - help.sigmacomputing.com/docs/use-sigma-mcp-server: https://help.sigmacomputing.com/docs/use-sigma-mcp-server (read 10 Nov 2026) - tableau.github.io/tableau-mcp/docs/intro: https://tableau.github.io/tableau-mcp/docs/intro (read 10 Nov 2026) - github.com/tableau/tableau-mcp/releases: https://github.com/tableau/tableau-mcp/releases (read 10 Nov 2026) - registry.npmjs.org/@tableau/mcp-server: https://registry.npmjs.org/@tableau/mcp-server (read 10 Nov 2026) - raw.githubusercontent.com/tableau/tableau-mcp/main/README.md: https://raw.githubusercontent.com/tableau/tableau-mcp/main/README.md (read 10 Nov 2026) - github.com/thoughtspot/mcp-server: https://github.com/thoughtspot/mcp-server (read 10 Nov 2026) - github.com/thoughtspot/mcp-server/releases: https://github.com/thoughtspot/mcp-server/releases (read 10 Nov 2026) - registry.npmjs.org/@thoughtspot/mcp-server: https://registry.npmjs.org/@thoughtspot/mcp-server (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/categories/bi/ --- # 8 MCP servers for catalogs and metadata, filed by data tool CATEGORY, read 10 Nov 2026 **The short answer.** For catalogs and metadata, the directory lists 8 MCP servers across 8 data tools: Alation, Atlan, Collibra, DataHub, Euno, OpenMetadata, Secoda, Select Star. 8 can run hosted and 4 locally; together they name 210 tools, read 10 Nov 2026. Every entry shows the same fourteen fields in the same order, from publisher and version to auth, access and context weight. Catalog and metadata servers describe data rather than query it: search, ownership, glossary terms, lineage and impact. Eight are listed, including the open-source DataHub and OpenMetadata servers and the hosted servers from Atlan, Euno and Select Star. | Data tool | Server | Kind | Transport | Access | Tools | Version | Last release | | --- | --- | --- | --- | --- | --- | --- | --- | | Alation | [Alation AI Agent SDK MCP server](/servers/alation-mcp/) | Official | Remote and local | Not stated in the source read | 12 | 1.0.0 | 26 Oct 2026 | | Atlan | [Atlan MCP Server](/servers/atlan-mcp/) | Official | Remote (hosted) | Configurable | 41 | Hosted at mcp.atlan.com/mcp (unversioned); local atlan-mcp-server 0.3.3 (deprecated) | 17 Feb 2026 | | Collibra | [Collibra MCP Server (chip)](/servers/collibra-mcp/) | Official | Remote and local | Configurable | 49 | V0.0.67 | 27 Oct 2026 | | DataHub | [DataHub MCP Server](/servers/datahub-mcp/) | Official | Remote and local | Configurable | 22 | 0.8.0 | 21 Oct 2026 | | Euno | [Euno MCP](/servers/euno-mcp/) | Official | Remote (hosted) | Configurable | 56 | Hosted service at api.app.euno.ai/mcp, no version published (docs read 10 Nov 2026) | none found | | OpenMetadata | [OpenMetadata MCP Server](/servers/openmetadata-mcp/) | Official | Remote (hosted) | Read and write | 15 | 2.0.7 (bundled with platform release 2.0.7-release; 2.1.0-rc1 also listed) | 27 Oct 2026 | | Secoda | [Secoda MCP](/servers/secoda-mcp/) | Official | Remote and local | Not stated in the source read | 8 | No releases; main branch last commit 26 Jan 2026 | none found | | Select Star | [Select Star MCP Server](/servers/select-star-mcp/) | Official | Remote (hosted) | Read only | 7 | Hosted, no version published | none found | Catalogs and metadata: 8 servers, filed by data tool, read 10 Nov 2026. **What Euno's server does not do.** Euno's server gives an agent governed context about the data; it does not run the query, which is the job of the warehouse's own server. ## In one sentence each - [Alation AI Agent SDK MCP server](/servers/alation-mcp/): Alation's server offers a lineage tool and a catalog search agent, and its remote form hides tools the current user cannot use. - [Atlan MCP Server](/servers/atlan-mcp/): Atlan's hosted per-tenant server runs as the signed-in user, scopes context by persona, offers an optional read-only mode and proposes writes for approval. - [Collibra MCP Server (chip)](/servers/collibra-mcp/): Collibra's server covers lineage with transformation logic, business terms and column semantics, classification, stewardship edits and data quality, under the connecting user's permissions. - [DataHub MCP Server](/servers/datahub-mcp/): DataHub's server returns column-level lineage with paths and SQL transformations, plus glossary, domain and owner discovery, with per-user tokens in HTTP mode. - [Euno MCP](/servers/euno-mcp/): Euno's hosted OAuth endpoint serves persona-scoped lineage, impact analysis, EQL search and catalog tools over its metadata graph, with no warehouse query tool. - [OpenMetadata MCP Server](/servers/openmetadata-mcp/): OpenMetadata's built-in server offers per-user OAuth, persona-scoped context, table and optional column lineage, and write tools for lineage, entities and tests. - [Secoda MCP](/servers/secoda-mcp/): Secoda's server combines catalog search, glossary, lineage and SQL over a workspace, as a hosted endpoint and an open-source wrapper. - [Select Star MCP Server](/servers/select-star-mcp/): Select Star's hosted, read-only server exposes catalog search, lineage, query history and semantic models with token setup. ## Other kinds of data tool - [Warehouses](/categories/warehouses/) - [Transformation](/categories/transformation/) - [Ingestion connectors](/categories/ingestion/) - [BI and analytics](/categories/bi/) - [Data observability](/categories/observability/) ## Sources - pypi.org/pypi/alation-ai-agent-mcp/json: https://pypi.org/pypi/alation-ai-agent-mcp/json (read 10 Nov 2026) - github.com/Alation/alation-ai-agent-sdk/releases: https://github.com/Alation/alation-ai-agent-sdk/releases (read 10 Nov 2026) - github.com/Alation/alation-ai-agent-sdk/blob/main/README.md: https://github.com/Alation/alation-ai-agent-sdk/blob/main/README.md (read 10 Nov 2026) - docs.atlan.com/product/capabilities/atlan-ai/references/mcp-security: https://docs.atlan.com/product/capabilities/atlan-ai/references/mcp-security (read 10 Nov 2026) - github.com/atlanhq/agent-toolkit/blob/main/modelcontextprotocol/README.md: https://github.com/atlanhq/agent-toolkit/blob/main/modelcontextprotocol/README.md (read 10 Nov 2026) - github.com/atlanhq/agent-toolkit/blob/main/CHANGELOG.md: https://github.com/atlanhq/agent-toolkit/blob/main/CHANGELOG.md (read 10 Nov 2026) - github.com/collibra/chip/blob/main/README.md: https://github.com/collibra/chip/blob/main/README.md (read 10 Nov 2026) - github.com/collibra/chip/releases: https://github.com/collibra/chip/releases (read 10 Nov 2026) - pypi.org/pypi/mcp-server-datahub/json: https://pypi.org/pypi/mcp-server-datahub/json (read 10 Nov 2026) - github.com/acryldata/mcp-server-datahub/releases/tag/v0.8.0: https://github.com/acryldata/mcp-server-datahub/releases/tag/v0.8.0 (read 10 Nov 2026) - docs.euno.ai/ai/mcp-setup.md: https://docs.euno.ai/ai/mcp-setup.md (read 10 Nov 2026) - docs.euno.ai/ai/mcp-setup/mcp-bridge-setup.md: https://docs.euno.ai/ai/mcp-setup/mcp-bridge-setup.md (read 10 Nov 2026) - docs.open-metadata.org/v2.0.x/how-to-guides/mcp.md: https://docs.open-metadata.org/v2.0.x/how-to-guides/mcp.md (read 10 Nov 2026) - github.com/open-metadata/OpenMetadata/releases/tag/2.0.7-release: https://github.com/open-metadata/OpenMetadata/releases/tag/2.0.7-release (read 10 Nov 2026) - docs.secoda.co/features/ai-assistant/secoda-mcp-server.md: https://docs.secoda.co/features/ai-assistant/secoda-mcp-server.md (read 10 Nov 2026) - github.com/secoda/secoda-mcp/releases: https://github.com/secoda/secoda-mcp/releases (read 10 Nov 2026) - api.github.com/repos/secoda/secoda-mcp/commits?per_page=1: https://api.github.com/repos/secoda/secoda-mcp/commits?per_page=1 (read 10 Nov 2026) - docs.selectstar.com/features/mcp-server: https://docs.selectstar.com/features/mcp-server (read 10 Nov 2026) - web.archive.org/web/20260614045618/https://docs.selectstar.com/features/mcp-server: https://web.archive.org/web/20260614045618/https://docs.selectstar.com/features/mcp-server (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/categories/catalogs/ --- # 3 MCP servers for data observability, filed by data tool CATEGORY, read 10 Nov 2026 **The short answer.** For data observability, the directory lists 3 MCP servers across 3 data tools: Monte Carlo, Sifflet, SYNQ. 2 can run hosted and 1 locally; together they name 159 tools, read 10 Nov 2026. Every entry shows the same fourteen fields in the same order, from publisher and version to auth, access and context weight. Data observability servers expose incidents, monitors and the assets they affect. Three are listed: Monte Carlo and SYNQ as hosted servers, and Sifflet as a local package. | Data tool | Server | Kind | Transport | Access | Tools | Version | Last release | | --- | --- | --- | --- | --- | --- | --- | --- | | Monte Carlo | [Monte Carlo MCP Server](/servers/monte-carlo-mcp/) | Official | Remote (hosted) | Configurable | 15 | Hosted at mcp.getmontecarlo.com/mcp, no version published | none found | | Sifflet | [Sifflet MCP Server](/servers/sifflet-mcp/) | Official | Local (stdio) | Read and write | 10 | 0.1.7 (PyPI); main at commit 1f390ca | 3 Aug 2026 | | SYNQ | [SYNQ MCP (Coalesce Quality)](/servers/synq-mcp/) | Official | Remote (hosted) | Configurable | 134 | Hosted, no version published | none found | Data observability: 3 servers, filed by data tool, read 10 Nov 2026. ## In one sentence each - [Monte Carlo MCP Server](/servers/monte-carlo-mcp/): Monte Carlo's hosted observability server uses per-user OAuth or scoped keys, a per-request read-only header and tool allow-lists. - [Sifflet MCP Server](/servers/sifflet-mcp/): Sifflet's local server covers asset search, incidents, monitors as code and downstream impact from one Sifflet API token. - [SYNQ MCP (Coalesce Quality)](/servers/synq-mcp/): SYNQ's hosted observability and lineage server uses per-user OAuth with tiered consent scopes, and tools hide when a scope is not granted. ## Other kinds of data tool - [Warehouses](/categories/warehouses/) - [Transformation](/categories/transformation/) - [Ingestion connectors](/categories/ingestion/) - [BI and analytics](/categories/bi/) - [Catalogs and metadata](/categories/catalogs/) ## Sources - docs.getmontecarlo.com/docs/ai-security: https://docs.getmontecarlo.com/docs/ai-security (read 10 Nov 2026) - docs.getmontecarlo.com/docs/mcp-server: https://docs.getmontecarlo.com/docs/mcp-server (read 10 Nov 2026) - raw.githubusercontent.com/siffletdata/sifflet-mcp/main/pyproject.toml: https://raw.githubusercontent.com/siffletdata/sifflet-mcp/main/pyproject.toml (read 10 Nov 2026) - pypi.org/project/sifflet-mcp: https://pypi.org/project/sifflet-mcp/ (read 10 Nov 2026) - docs.siffletdata.com/docs/sifflet-mcp-server.md: https://docs.siffletdata.com/docs/sifflet-mcp-server.md (read 10 Nov 2026) - docs.synq.io/scout/mcp.md: https://docs.synq.io/scout/mcp.md (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/categories/observability/ --- # Snowflake Labs MCP Server and Snowflake-managed MCP server, field by field: 6 and 6 tools named SIDE BY SIDE, read 10 Nov 2026 **The short answer.** Snowflake Labs MCP Server (Snowflake Labs, remote and local) names 6 tools; Snowflake-managed MCP server (Snowflake, remote (hosted)) names 6. Versions read on 10 Nov 2026: 1.4.2 and no version number published. The page puts the same fields in the same rows and draws no conclusion. | Field | Snowflake Labs MCP Server | Snowflake-managed MCP server | | --- | --- | --- | | Data tool | [Snowflake](/mcp/snowflake/) | [Snowflake](/mcp/snowflake/) | | Publisher | Snowflake Labs | Snowflake | | Kind | Official, Labs org | Official | | Repository or docs | [github.com/Snowflake-Labs/mcp](https://github.com/Snowflake-Labs/mcp) | [docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-mcp](https://docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-mcp) | | Version or commit | 1.4.2 | Docs page, MCP revision 25 Nov 2025 | | Last release | 15 May 2026 | none found on 10 Nov 2026 | | Transport | Remote and local | Remote (hosted) | | Auth | Snowflake Python connector: password/PAT, key pair, OAuth, SSO, MFA; honors the connection role's RBAC. | Snowflake OAuth (default) or External OAuth; PAT supported; RBAC per MCP server and per tool. | | Access | Configurable | Configurable | | Tools listed | 6 named in the evidence | 6 named in the evidence | | Tool definitions read at | [raw.githubusercontent.com/Snowflake-Labs/mcp/main/README.md](https://raw.githubusercontent.com/Snowflake-Labs/mcp/main/README.md) | [docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-mcp](https://docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-mcp) | | Context weight | Not measured: The Snowflake Labs server, marked deprecated in its README, which describes config-driven tool groups (Cortex Search, Analyst, Agent, object management, SQL execution, semantic views) but publishes no per-tool names or descriptions; facts.json lists these six groups. | Not measured: The docs page defines user-configured tool types (CORTEX_AGENT_RUN, CODING_AGENT, CORTEX_SEARCH_SERVICE_QUERY, CORTEX_ANALYST_MESSAGE, SYSTEM_EXECUTE_SQL, GENERIC) used to build tools in a CREATE MCP SERVER specification; it publishes no fixed list of named tools, so facts.json lists tool types not tools. | | Status | Exists, but the README states it is deprecated and no longer maintained and points to the Snowflake-managed MCP server (read 10 Nov 2026). | Documented by Snowflake as the Snowflake-managed MCP server (read 10 Nov 2026); the Snowflake-Labs repository points users here. | | Read | 10 Nov 2026 | 10 Nov 2026 | The same fourteen fields for both servers, read 10 Nov 2026. ## In one sentence each - [Snowflake Labs MCP Server](/servers/snowflake-labs-mcp/): The Labs server gives config-driven access to Cortex services, object management and semantic views under the connecting role; its README marks it deprecated. - [Snowflake-managed MCP server](/servers/snowflake-managed-mcp/): Snowflake's hosted endpoint applies RBAC per server and per tool with OAuth, and serves Cortex Analyst over semantic views, Cortex Search and SQL. ## The read facts, topic by topic | Topic | Snowflake Labs MCP Server | Snowflake-managed MCP server | | --- | --- | --- | | Publisher and releases | Snowflake Labs marks its Snowflake-Labs/mcp repository as deprecated and unmaintained, pointing users to the official Snowflake MCP server. The latest PyPI/GitHub release of snowflake-labs-mcp is v1.4.2, published 15 May 2026 and titled as the release that starts deprecation. PyPI lists snowflake-labs-mcp 1.4.2 uploaded 15 May 2026 by Snowflake Labs authors. | Snowflake documents a Snowflake-managed MCP server that lets agents retrieve data from Snowflake accounts without deploying separate infrastructure. The docs state support for MCP protocol revision 25 Nov 2025; no semantic version number or release date is published on the page. | | Clients, auth and scoping | The server served Cortex Search, Cortex Analyst and Cortex Agent services to MCP clients as tools, configured in a YAML file. The server honors the RBAC grants of the role used for the connection, so access is scoped to that Snowflake role. | MCP clients discover and invoke Cortex Analyst, Cortex Search and Cortex Agents as tools on the managed server. Access to the MCP server does not give access to tools; privileges are granted per tool, so results depend on the calling role. | | Access and safety | SQL statement types are allowed or denied per type in a sql_statement_permissions config block. Authentication supports password, key pair, OAuth, SSO and MFA via the Snowflake connector. | Authentication is Snowflake OAuth by default, with optional External OAuth, and role-based access control on the server and tools. Docs list distinct privileges: USAGE on the MCP server plus SELECT on the semantic view, USAGE on the Cortex Search service, agent, or UDF. Docs warn that tool annotations are client hints and not security controls; RBAC is the control. The managed server is available in government regions except Azure US Gov Virginia and has not completed formal FedRAMP assessment. | | What it reaches | Covers Snowflake only: Cortex Search, Analyst, Agent, object management and SQL execution. | Serves Snowflake data only; custom tools can wrap UDFs and stored procedures. | | Lineage and impact | Legacy tooling included semantic view discovery and querying plus Cortex Analyst, but no lineage tool is listed. None found: A lineage or column-lineage tool in the Snowflake-Labs/mcp README (checked tool and capability list). | Tool types are Cortex Agent, coding agent, Cortex Search, Cortex Analyst, SQL execution and generic UDF/procedure tools; none is a lineage tool. None found: A lineage, column-lineage or object-dependency tool type on the Snowflake-managed MCP server page (tool types list read). | | Definitions, ownership and rules | The legacy server could discover Snowflake Semantic Views and their components, which carry business definitions. | Cortex Analyst tools use semantic views, which hold business definitions, and the server accepts semantic views only, not semantic models. | | Freshness and upkeep | Queries run live against Snowflake through the Python connector at call time; no collected or cached metadata store is described. | Tools run against live Snowflake objects; no metadata snapshot or refresh schedule is described. | | Profile | Transports are stdio (default), sse (legacy) and streamable-http; runs via uvx or Docker. | The server is a Snowflake object created with CREATE MCP SERVER, reachable at an account-scoped HTTP URL (remote). | Each cell joins the facts read for that topic; the full list with sources is on each entry. Read 10 Nov 2026. ## Tool names, as published ### Snowflake Labs MCP Server `Cortex Search services`, `Cortex Analyst services`, `Cortex Agent services`, `object manager group`, `query manager (SQL execution) group`, `semantic manager group` ### Snowflake-managed MCP server `CORTEX_AGENT_RUN`, `CODING_AGENT`, `CORTEX_SEARCH_SERVICE_QUERY`, `CORTEX_ANALYST_MESSAGE`, `SYSTEM_EXECUTE_SQL`, `GENERIC` ## Read next - [Snowflake Labs MCP Server, the full entry](/servers/snowflake-labs-mcp/) - [Snowflake-managed MCP server, the full entry](/servers/snowflake-managed-mcp/) - [Every side-by-side](/compare/) ## Sources - raw.githubusercontent.com/Snowflake-Labs/mcp/main/README.md: https://raw.githubusercontent.com/Snowflake-Labs/mcp/main/README.md (read 10 Nov 2026) - github.com/Snowflake-Labs/mcp/releases: https://github.com/Snowflake-Labs/mcp/releases (read 10 Nov 2026) - pypi.org/pypi/snowflake-labs-mcp/json: https://pypi.org/pypi/snowflake-labs-mcp/json (read 10 Nov 2026) - docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-mcp: https://docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-mcp (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/compare/snowflake-labs-mcp-vs-snowflake-managed-mcp/ --- # mcp-snowflake-server (community) and Snowflake-managed MCP server, field by field: 8 and 6 tools named SIDE BY SIDE, read 10 Nov 2026 **The short answer.** mcp-snowflake-server (community) (isaacwasserman, local (stdio)) names 8 tools; Snowflake-managed MCP server (Snowflake, remote (hosted)) names 6. Versions read on 10 Nov 2026: 0.4.0 and no version number published. The page puts the same fields in the same rows and draws no conclusion. | Field | mcp-snowflake-server (community) | Snowflake-managed MCP server | | --- | --- | --- | | Data tool | [Snowflake](/mcp/snowflake/) | [Snowflake](/mcp/snowflake/) | | Publisher | isaacwasserman (community) | Snowflake | | Kind | Community | Official | | Repository or docs | [github.com/isaacwasserman/mcp-snowflake-server](https://github.com/isaacwasserman/mcp-snowflake-server) | [docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-mcp](https://docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-mcp) | | Version or commit | 0.4.0 | Docs page, MCP revision 25 Nov 2025 | | Last release | 10 Apr 2025 | none found on 10 Nov 2026 | | Transport | Local (stdio) | Remote (hosted) | | Auth | Snowflake user credentials (password, private key, external browser) or connections.toml. | Snowflake OAuth (default) or External OAuth; PAT supported; RBAC per MCP server and per tool. | | Access | Configurable | Configurable | | Tools listed | 8 named in the evidence | 6 named in the evidence | | Tool definitions read at | [raw.githubusercontent.com/isaacwasserman/mcp-snowflake-server/main/README.md](https://raw.githubusercontent.com/isaacwasserman/mcp-snowflake-server/main/README.md) | [docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-mcp](https://docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-mcp) | | Context weight | 103 tokens for 8 tool definitions (8 with a description), measured from the README read | Not measured: The docs page defines user-configured tool types (CORTEX_AGENT_RUN, CODING_AGENT, CORTEX_SEARCH_SERVICE_QUERY, CORTEX_ANALYST_MESSAGE, SYSTEM_EXECUTE_SQL, GENERIC) used to build tools in a CREATE MCP SERVER specification; it publishes no fixed list of named tools, so facts.json lists tool types not tools. | | Status | Exists as a community server; PyPI 0.4.0 dated 10 Apr 2025, GitHub pushed 7 Oct 2025 (GitHub API). | Documented by Snowflake as the Snowflake-managed MCP server (read 10 Nov 2026); the Snowflake-Labs repository points users here. | | Read | 10 Nov 2026 | 10 Nov 2026 | The same fourteen fields for both servers, read 10 Nov 2026. ## In one sentence each - [mcp-snowflake-server (community)](/servers/isaacwasserman-mcp-snowflake-server/): A community server for schema browsing and read-only SQL on Snowflake, with write tools off by default. - [Snowflake-managed MCP server](/servers/snowflake-managed-mcp/): Snowflake's hosted endpoint applies RBAC per server and per tool with OAuth, and serves Cortex Analyst over semantic views, Cortex Search and SQL. ## The read facts, topic by topic | Topic | mcp-snowflake-server (community) | Snowflake-managed MCP server | | --- | --- | --- | | Publisher and releases | The server is a community project by isaacwasserman, not published by Snowflake; license is GPL-3.0 per GitHub API. PyPI lists mcp-snowflake-server 0.4.0 uploaded 10 Apr 2025; the GitHub repo was last pushed 7 Oct 2025. | Snowflake documents a Snowflake-managed MCP server that lets agents retrieve data from Snowflake accounts without deploying separate infrastructure. The docs state support for MCP protocol revision 25 Nov 2025; no semantic version number or release date is published on the page. | | Clients, auth and scoping | The server exposes a memo://insights resource and an append_insight tool that stores agent-discovered insights. When prefetch is enabled it exposes per-table schema summaries as resources. | MCP clients discover and invoke Cortex Analyst, Cortex Search and Cortex Agents as tools on the managed server. Access to the MCP server does not give access to tools; privileges are granted per tool, so results depend on the calling role. | | Access and safety | Write operations are disabled by default and enabled only with --allow-write. Tool exclusion and database/schema/table exclusion patterns are supported. Authentication uses Snowflake user credentials, a private key path or external browser; the README shows plain passwords in config. | Authentication is Snowflake OAuth by default, with optional External OAuth, and role-based access control on the server and tools. Docs list distinct privileges: USAGE on the MCP server plus SELECT on the semantic view, USAGE on the Cortex Search service, agent, or UDF. Docs warn that tool annotations are client hints and not security controls; RBAC is the control. The managed server is available in government regions except Azure US Gov Virginia and has not completed formal FedRAMP assessment. | | What it reaches | Covers Snowflake only. | Serves Snowflake data only; custom tools can wrap UDFs and stored procedures. | | Lineage and impact | Schema tools list databases, schemas, tables and describe columns; no lineage tool is listed. None found: A lineage or dependency tool in the README tool list. | Tool types are Cortex Agent, coding agent, Cortex Search, Cortex Analyst, SQL execution and generic UDF/procedure tools; none is a lineage tool. None found: A lineage, column-lineage or object-dependency tool type on the Snowflake-managed MCP server page (tool types list read). | | Definitions, ownership and rules | Describe_table returns column names, types, nullability, defaults and comments, so table and column comments reach the agent. | Cortex Analyst tools use semantic views, which hold business definitions, and the server accepts semantic views only, not semantic models. | | Freshness and upkeep | Metadata is read live from Snowflake on each tool call; no scheduled collection is described. | Tools run against live Snowflake objects; no metadata snapshot or refresh schedule is described. | | Profile | Run via uvx or uv with stdio for Claude Desktop; install via Smithery is also documented. | The server is a Snowflake object created with CREATE MCP SERVER, reachable at an account-scoped HTTP URL (remote). | Each cell joins the facts read for that topic; the full list with sources is on each entry. Read 10 Nov 2026. ## Tool names, as published ### mcp-snowflake-server (community) `read_query`, `write_query`, `create_table`, `list_databases`, `list_schemas`, `list_tables`, `describe_table`, `append_insight` ### Snowflake-managed MCP server `CORTEX_AGENT_RUN`, `CODING_AGENT`, `CORTEX_SEARCH_SERVICE_QUERY`, `CORTEX_ANALYST_MESSAGE`, `SYSTEM_EXECUTE_SQL`, `GENERIC` ## Read next - [mcp-snowflake-server (community), the full entry](/servers/isaacwasserman-mcp-snowflake-server/) - [Snowflake-managed MCP server, the full entry](/servers/snowflake-managed-mcp/) - [Every side-by-side](/compare/) ## Sources - raw.githubusercontent.com/isaacwasserman/mcp-snowflake-server/main/README.md: https://raw.githubusercontent.com/isaacwasserman/mcp-snowflake-server/main/README.md (read 10 Nov 2026) - pypi.org/pypi/mcp-snowflake-server/json: https://pypi.org/pypi/mcp-snowflake-server/json (read 10 Nov 2026) - docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-mcp: https://docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-mcp (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/compare/isaacwasserman-mcp-snowflake-server-vs-snowflake-managed-mcp/ --- # mcp-snowflake-server (community) and Snowflake Labs MCP Server, field by field: 8 and 6 tools named SIDE BY SIDE, read 10 Nov 2026 **The short answer.** mcp-snowflake-server (community) (isaacwasserman, local (stdio)) names 8 tools; Snowflake Labs MCP Server (Snowflake Labs, remote and local) names 6. Versions read on 10 Nov 2026: 0.4.0 and 1.4.2. The page puts the same fields in the same rows and draws no conclusion. | Field | mcp-snowflake-server (community) | Snowflake Labs MCP Server | | --- | --- | --- | | Data tool | [Snowflake](/mcp/snowflake/) | [Snowflake](/mcp/snowflake/) | | Publisher | isaacwasserman (community) | Snowflake Labs | | Kind | Community | Official, Labs org | | Repository or docs | [github.com/isaacwasserman/mcp-snowflake-server](https://github.com/isaacwasserman/mcp-snowflake-server) | [github.com/Snowflake-Labs/mcp](https://github.com/Snowflake-Labs/mcp) | | Version or commit | 0.4.0 | 1.4.2 | | Last release | 10 Apr 2025 | 15 May 2026 | | Transport | Local (stdio) | Remote and local | | Auth | Snowflake user credentials (password, private key, external browser) or connections.toml. | Snowflake Python connector: password/PAT, key pair, OAuth, SSO, MFA; honors the connection role's RBAC. | | Access | Configurable | Configurable | | Tools listed | 8 named in the evidence | 6 named in the evidence | | Tool definitions read at | [raw.githubusercontent.com/isaacwasserman/mcp-snowflake-server/main/README.md](https://raw.githubusercontent.com/isaacwasserman/mcp-snowflake-server/main/README.md) | [raw.githubusercontent.com/Snowflake-Labs/mcp/main/README.md](https://raw.githubusercontent.com/Snowflake-Labs/mcp/main/README.md) | | Context weight | 103 tokens for 8 tool definitions (8 with a description), measured from the README read | Not measured: The Snowflake Labs server, marked deprecated in its README, which describes config-driven tool groups (Cortex Search, Analyst, Agent, object management, SQL execution, semantic views) but publishes no per-tool names or descriptions; facts.json lists these six groups. | | Status | Exists as a community server; PyPI 0.4.0 dated 10 Apr 2025, GitHub pushed 7 Oct 2025 (GitHub API). | Exists, but the README states it is deprecated and no longer maintained and points to the Snowflake-managed MCP server (read 10 Nov 2026). | | Read | 10 Nov 2026 | 10 Nov 2026 | The same fourteen fields for both servers, read 10 Nov 2026. ## In one sentence each - [mcp-snowflake-server (community)](/servers/isaacwasserman-mcp-snowflake-server/): A community server for schema browsing and read-only SQL on Snowflake, with write tools off by default. - [Snowflake Labs MCP Server](/servers/snowflake-labs-mcp/): The Labs server gives config-driven access to Cortex services, object management and semantic views under the connecting role; its README marks it deprecated. ## The read facts, topic by topic | Topic | mcp-snowflake-server (community) | Snowflake Labs MCP Server | | --- | --- | --- | | Publisher and releases | The server is a community project by isaacwasserman, not published by Snowflake; license is GPL-3.0 per GitHub API. PyPI lists mcp-snowflake-server 0.4.0 uploaded 10 Apr 2025; the GitHub repo was last pushed 7 Oct 2025. | Snowflake Labs marks its Snowflake-Labs/mcp repository as deprecated and unmaintained, pointing users to the official Snowflake MCP server. The latest PyPI/GitHub release of snowflake-labs-mcp is v1.4.2, published 15 May 2026 and titled as the release that starts deprecation. PyPI lists snowflake-labs-mcp 1.4.2 uploaded 15 May 2026 by Snowflake Labs authors. | | Clients, auth and scoping | The server exposes a memo://insights resource and an append_insight tool that stores agent-discovered insights. When prefetch is enabled it exposes per-table schema summaries as resources. | The server served Cortex Search, Cortex Analyst and Cortex Agent services to MCP clients as tools, configured in a YAML file. The server honors the RBAC grants of the role used for the connection, so access is scoped to that Snowflake role. | | Access and safety | Write operations are disabled by default and enabled only with --allow-write. Tool exclusion and database/schema/table exclusion patterns are supported. Authentication uses Snowflake user credentials, a private key path or external browser; the README shows plain passwords in config. | SQL statement types are allowed or denied per type in a sql_statement_permissions config block. Authentication supports password, key pair, OAuth, SSO and MFA via the Snowflake connector. | | What it reaches | Covers Snowflake only. | Covers Snowflake only: Cortex Search, Analyst, Agent, object management and SQL execution. | | Lineage and impact | Schema tools list databases, schemas, tables and describe columns; no lineage tool is listed. None found: A lineage or dependency tool in the README tool list. | Legacy tooling included semantic view discovery and querying plus Cortex Analyst, but no lineage tool is listed. None found: A lineage or column-lineage tool in the Snowflake-Labs/mcp README (checked tool and capability list). | | Definitions, ownership and rules | Describe_table returns column names, types, nullability, defaults and comments, so table and column comments reach the agent. | The legacy server could discover Snowflake Semantic Views and their components, which carry business definitions. | | Freshness and upkeep | Metadata is read live from Snowflake on each tool call; no scheduled collection is described. | Queries run live against Snowflake through the Python connector at call time; no collected or cached metadata store is described. | | Profile | Run via uvx or uv with stdio for Claude Desktop; install via Smithery is also documented. | Transports are stdio (default), sse (legacy) and streamable-http; runs via uvx or Docker. | Each cell joins the facts read for that topic; the full list with sources is on each entry. Read 10 Nov 2026. ## Tool names, as published ### mcp-snowflake-server (community) `read_query`, `write_query`, `create_table`, `list_databases`, `list_schemas`, `list_tables`, `describe_table`, `append_insight` ### Snowflake Labs MCP Server `Cortex Search services`, `Cortex Analyst services`, `Cortex Agent services`, `object manager group`, `query manager (SQL execution) group`, `semantic manager group` ## Read next - [mcp-snowflake-server (community), the full entry](/servers/isaacwasserman-mcp-snowflake-server/) - [Snowflake Labs MCP Server, the full entry](/servers/snowflake-labs-mcp/) - [Every side-by-side](/compare/) ## Sources - raw.githubusercontent.com/isaacwasserman/mcp-snowflake-server/main/README.md: https://raw.githubusercontent.com/isaacwasserman/mcp-snowflake-server/main/README.md (read 10 Nov 2026) - pypi.org/pypi/mcp-snowflake-server/json: https://pypi.org/pypi/mcp-snowflake-server/json (read 10 Nov 2026) - raw.githubusercontent.com/Snowflake-Labs/mcp/main/README.md: https://raw.githubusercontent.com/Snowflake-Labs/mcp/main/README.md (read 10 Nov 2026) - github.com/Snowflake-Labs/mcp/releases: https://github.com/Snowflake-Labs/mcp/releases (read 10 Nov 2026) - pypi.org/pypi/snowflake-labs-mcp/json: https://pypi.org/pypi/snowflake-labs-mcp/json (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/compare/isaacwasserman-mcp-snowflake-server-vs-snowflake-labs-mcp/ --- # BigQuery remote MCP server and MCP Toolbox for Databases (BigQuery tools), field by field: 9 and 9 tools named SIDE BY SIDE, read 10 Nov 2026 **The short answer.** BigQuery remote MCP server (Google Cloud, remote (hosted)) names 9 tools; MCP Toolbox for Databases (BigQuery tools) (Google, remote and local) names 9. Versions read on 10 Nov 2026: no version number published and v1.16.0. The page puts the same fields in the same rows and draws no conclusion. | Field | BigQuery remote MCP server | MCP Toolbox for Databases (BigQuery tools) | | --- | --- | --- | | Data tool | [BigQuery](/mcp/bigquery/) | [BigQuery](/mcp/bigquery/) | | Publisher | Google Cloud | Google (googleapis) | | Kind | Official | Official | | Repository or docs | [docs.cloud.google.com/bigquery/docs/use-bigquery-mcp](https://docs.cloud.google.com/bigquery/docs/use-bigquery-mcp) | [github.com/googleapis/mcp-toolbox](https://github.com/googleapis/mcp-toolbox) | | Version or commit | Docs reference last updated 3 Nov 2026 | V1.16.0 | | Last release | none found on 10 Nov 2026 | 5 Nov 2026 | | Transport | Remote (hosted) | Remote and local | | Auth | OAuth 2.0 with IAM (Google Cloud identities, OAuth client, or agent identity). | Application Default Credentials, client OAuth token forwarding, or service account impersonation. | | Access | Configurable | Configurable | | Tools listed | 9 named in the evidence | 9 named in the evidence | | Tool definitions read at | [docs.cloud.google.com/bigquery/docs/reference/mcp](https://docs.cloud.google.com/bigquery/docs/reference/mcp) | [raw.githubusercontent.com/googleapis/mcp-toolbox/main/docs/en/integrations/bigquery/prebuilt-configs/bigquery.md](https://raw.githubusercontent.com/googleapis/mcp-toolbox/main/docs/en/integrations/bigquery/prebuilt-configs/bigquery.md) | | Context weight | 1,487 tokens for 9 tool definitions (9 with a description), measured from the docs page read | 156 tokens for 9 tool definitions (9 with a description), measured from the docs page read | | Status | Exists; Google-managed endpoint, reference page last updated 3 Nov 2026 (read 10 Nov 2026). | Repo googleapis/genai-toolbox now lives at googleapis/mcp-toolbox (read 10 Nov 2026); BigQuery tools documented there. | | Read | 10 Nov 2026 | 10 Nov 2026 | The same fourteen fields for both servers, read 10 Nov 2026. ## In one sentence each - [BigQuery remote MCP server](/servers/bigquery-remote-mcp/): Google Cloud's managed endpoint gives IAM-scoped access to a SELECT-only query tool and dataset and table metadata tools. - [MCP Toolbox for Databases (BigQuery tools)](/servers/mcp-toolbox-bigquery/): Google's open-source Toolbox offers BigQuery SQL and metadata tools with a read-only mode, dataset allowlists and Knowledge Catalog search. ## The read facts, topic by topic | Topic | BigQuery remote MCP server | MCP Toolbox for Databases (BigQuery tools) | | --- | --- | --- | | Publisher and releases | Google documents a BigQuery remote MCP server, enabled when the BigQuery API is enabled. The reference page for bigquery.googleapis.com MCP shows Last updated 3 Nov 2026 UTC; no semantic version is published. | MCP Toolbox for Databases is an open source MCP server in the googleapis organization (formerly genai-toolbox) with a BigQuery prebuilt config. The latest GitHub release is v1.16.0, whose release timestamp on the page is 5 Nov 2026. | | Clients, auth and scoping | The server can run queries, get metadata and list resources for AI apps including Claude and ChatGPT. OAuth 2.0 with IAM is used so access follows the authenticated Google Cloud identity. | Tools include ask_data_insights (Conversational Analytics API), forecast and analyze_contribution. With useClientOAuth the client's OAuth token is forwarded so calls run as the asking user. | | Access and safety | Execute_sql_readonly is restricted to SELECT statements and is preferred over execute_sql. Execute_sql can run INSERT, UPDATE, DELETE and CREATE statements. Google remote MCP servers offer centralized audit logging and optional Model Armor protection. | Source options readOnly and writeMode (allowed, blocked, protected) enforce read-only operation; blocked allows only SELECT. AllowedDatasets restricts tools to listed datasets and impersonateServiceAccount supports service account impersonation. BIGQUERY_MAXIMUM_BYTES_BILLED caps per-query bytes scanned. | | What it reaches | Covers BigQuery and BigLake tables and namespaces. | The BigQuery source covers BigQuery; the Toolbox itself supports other databases. | | Lineage and impact | Tools cover dataset and table metadata and SQL execution; no lineage tool among the nine listed. None found: A lineage or Dataplex lineage tool in the BigQuery MCP tools reference (nine tools listed). | The BigQuery prebuilt includes search_catalog which searches Knowledge Catalog (formerly Dataplex) entries; no lineage tool is listed. None found: A lineage tool in the BigQuery prebuilt tool list. | | Definitions, ownership and rules | Get_table_info and get_dataset_info return metadata for BigQuery and BigLake objects; no glossary or policy tool is listed. | Search_catalog finds tables, views, models, routines and connections via Knowledge Catalog metadata. | | Freshness and upkeep | Calls hit the live BigQuery APIs; the server has no quotas of its own beyond the underlying API quotas. | Tools query BigQuery and Knowledge Catalog live through ADC or OAuth; no collection mechanism. | | Profile | Global endpoint bigquery.googleapis.com/mcp with Streamable HTTP transport. | The README shows running with a --stdio flag for MCP clients. | Each cell joins the facts read for that topic; the full list with sources is on each entry. Read 10 Nov 2026. ## Tool names, as published ### BigQuery remote MCP server `list_dataset_ids`, `get_dataset_info`, `list_table_ids`, `get_table_info`, `execute_sql_readonly`, `execute_sql`, `get_query_results`, `cancel_job`, `get_job` ### MCP Toolbox for Databases (BigQuery tools) `analyze_contribution`, `ask_data_insights`, `execute_sql`, `forecast`, `get_dataset_info`, `get_table_info`, `list_dataset_ids`, `list_table_ids`, `search_catalog` ## Read next - [BigQuery remote MCP server, the full entry](/servers/bigquery-remote-mcp/) - [MCP Toolbox for Databases (BigQuery tools), the full entry](/servers/mcp-toolbox-bigquery/) - [Every side-by-side](/compare/) ## Sources - docs.cloud.google.com/bigquery/docs/use-bigquery-mcp: https://docs.cloud.google.com/bigquery/docs/use-bigquery-mcp (read 10 Nov 2026) - docs.cloud.google.com/bigquery/docs/reference/mcp: https://docs.cloud.google.com/bigquery/docs/reference/mcp (read 10 Nov 2026) - raw.githubusercontent.com/googleapis/mcp-toolbox/main/README.md: https://raw.githubusercontent.com/googleapis/mcp-toolbox/main/README.md (read 10 Nov 2026) - github.com/googleapis/mcp-toolbox/releases: https://github.com/googleapis/mcp-toolbox/releases (read 10 Nov 2026) - raw.githubusercontent.com/googleapis/mcp-toolbox/main/docs/en/integrations/bigquery/tools/bigquery-search-catalog.md: https://raw.githubusercontent.com/googleapis/mcp-toolbox/main/docs/en/integrations/bigquery/tools/bigquery-search-catalog.md (read 10 Nov 2026) - raw.githubusercontent.com/googleapis/mcp-toolbox/main/docs/en/integrations/bigquery/prebuilt-configs/bigquery.md: https://raw.githubusercontent.com/googleapis/mcp-toolbox/main/docs/en/integrations/bigquery/prebuilt-configs/bigquery.md (read 10 Nov 2026) - raw.githubusercontent.com/googleapis/mcp-toolbox/main/docs/en/integrations/bigquery/source.md: https://raw.githubusercontent.com/googleapis/mcp-toolbox/main/docs/en/integrations/bigquery/source.md (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/compare/bigquery-remote-mcp-vs-mcp-toolbox-bigquery/ --- # Databricks Labs MCP servers and Databricks managed MCP servers, field by field: 3 and 4 tools named SIDE BY SIDE, read 10 Nov 2026 **The short answer.** Databricks Labs MCP servers (Databricks Labs, remote and local) names 3 tools; Databricks managed MCP servers (Databricks, remote (hosted)) names 4. Versions read on 10 Nov 2026: no version number published and no version number published. The page puts the same fields in the same rows and draws no conclusion. | Field | Databricks Labs MCP servers | Databricks managed MCP servers | | --- | --- | --- | | Data tool | [Databricks](/mcp/databricks/) | [Databricks](/mcp/databricks/) | | Publisher | Databricks Labs | Databricks | | Kind | Official, Labs org | Official | | Repository or docs | [github.com/databrickslabs/mcp](https://github.com/databrickslabs/mcp) | [docs.databricks.com/aws/en/generative-ai/mcp/managed-mcp](https://docs.databricks.com/aws/en/generative-ai/mcp/managed-mcp) | | Version or commit | none found on 10 Nov 2026 | Docs page last updated Nov 9, 2026 | | Last release | none found on 10 Nov 2026 | none found on 10 Nov 2026 | | Transport | Remote and local | Remote (hosted) | | Auth | Databricks CLI-style credentials; Unity Catalog permissions. | Databricks workspace sign-in; caller's Unity Catalog permissions; Unity Gateway grants and policies. | | Access | Not stated in the source read | Configurable | | Tools listed | 3 named in the evidence | 4 named in the evidence | | Tool definitions read at | [raw.githubusercontent.com/databrickslabs/mcp/master/README.md](https://raw.githubusercontent.com/databrickslabs/mcp/master/README.md) | [docs.databricks.com/aws/en/generative-ai/mcp/managed-mcp](https://docs.databricks.com/aws/en/generative-ai/mcp/managed-mcp) | | Context weight | Not measured: README says the tool list is inferred at startup from the Unity Catalog functions, vector search indexes and Genie spaces the user configures, so no fixed per-tool list or descriptions are published. | 94 tokens for 4 tool definitions (4 with a description), measured from the docs page read | | Status | Exists; README marks Unity Catalog server deprecated in favor of managed MCP servers; last push 26 Oct 2026 (GitHub API); no tagged release read. | Documented by Databricks as Databricks-provided MCPs (read 10 Nov 2026); no versioned release stream found. | | Read | 10 Nov 2026 | 10 Nov 2026 | The same fourteen fields for both servers, read 10 Nov 2026. ## In one sentence each - [Databricks Labs MCP servers](/servers/databrickslabs-mcp/): The Labs server exposes Unity Catalog functions, vector search indexes and Genie spaces as tools; its README marks it deprecated in favor of the managed servers. - [Databricks managed MCP servers](/servers/databricks-managed-mcp/): Databricks-provided MCP endpoints run under the caller's Unity Catalog grants and include Genie spaces for questions over a governed semantic layer. ## The read facts, topic by topic | Topic | Databricks Labs MCP servers | Databricks managed MCP servers | | --- | --- | --- | | Publisher and releases | Databrickslabs/mcp is a Databricks Labs repository whose README calls the Unity Catalog server deprecated in favor of Managed MCP servers. The repo has no GitHub release entries read; GitHub API shows last push 26 Oct 2026 and no archive flag. | Databricks documents Databricks-provided MCPs in the system.ai schema, with no server to host or register; the page was last updated Nov 9, 2026. Genie One MCP is generally available while dbsql, sandbox and web_search require the Unity Gateway beta. | | Clients, auth and scoping | Tools are inferred at startup from the functions and vector search indexes in a given Unity Catalog schema plus listed Genie spaces. | System.ai.genie_one_mcp answers natural-language analytics questions grounded in Genie Ontology. The dbsql MCP runs SQL using the caller's Unity Catalog and warehouse permissions, so results are scoped to the asking user. | | Access and safety | Databricks managed MCP note in the README states permissions are always enforced for managed servers. | Each system.ai MCP is a Unity Catalog securable governed by grants and per-call policies via Unity Gateway. Genie Agent MCP calls bypass Unity Gateway and rely on permissions of the Genie Agent and its data. | | What it reaches | Covers Databricks Unity Catalog objects: functions, vector search indexes, Genie spaces. | Beyond data, Databricks ships MCPs for Slack, GitHub, Atlassian, Google Drive, Gmail and Microsoft 365. | | Lineage and impact | The Unity Catalog server exposes UC functions, vector search indexes and Genie spaces as tools; no lineage tool. None found: A lineage tool in the README supported-tools list. | The docs list Genie One, dbsql, web_search and sandbox MCPs; no lineage-specific MCP is listed. None found: A lineage or column-lineage MCP on the Databricks-provided MCPs page (list read). | | Definitions, ownership and rules | Genie spaces and UC functions carry curated descriptions to agents; no glossary or classification tool is listed. | Genie One resolves business terms, metric definitions and table relationships through Genie Ontology, described as a governed semantic layer. | | Freshness and upkeep | Tool list is built once at startup from the schema; no continuous refresh described. | Calls query the workspace live; no collection or refresh mechanism is described on this page. | | Profile | Developer tools server in the same repo is marked under construction and not yet usable. Local use is via stdio and Databricks Apps deployment uses Streamable HTTP. | Endpoint pattern is /ai-gateway/mcp-services/. | Each cell joins the facts read for that topic; the full list with sources is on each entry. Read 10 Nov 2026. ## Tool names, as published ### Databricks Labs MCP servers `UC functions (one tool per function)`, `vector search index tools`, `Genie space tools` ### Databricks managed MCP servers `system.ai.genie_one_mcp`, `system.ai.dbsql`, `system.ai.web_search`, `system.ai.sandbox` ## Read next - [Databricks Labs MCP servers, the full entry](/servers/databrickslabs-mcp/) - [Databricks managed MCP servers, the full entry](/servers/databricks-managed-mcp/) - [Every side-by-side](/compare/) ## Sources - raw.githubusercontent.com/databrickslabs/mcp/master/README.md: https://raw.githubusercontent.com/databrickslabs/mcp/master/README.md (read 10 Nov 2026) - docs.databricks.com/aws/en/generative-ai/mcp/managed-mcp: https://docs.databricks.com/aws/en/generative-ai/mcp/managed-mcp (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/compare/databrickslabs-mcp-vs-databricks-managed-mcp/ --- # Atlan MCP Server and DataHub MCP Server, field by field: 41 and 22 tools named SIDE BY SIDE, read 10 Nov 2026 **The short answer.** Atlan MCP Server (Atlan, remote (hosted)) names 41 tools; DataHub MCP Server (Acryl Data, remote and local) names 22. Versions read on 10 Nov 2026: 0.3.3 and 0.8.0. The page puts the same fields in the same rows and draws no conclusion. | Field | Atlan MCP Server | DataHub MCP Server | | --- | --- | --- | | Data tool | [Atlan](/mcp/atlan/) | [DataHub](/mcp/datahub/) | | Publisher | Atlan | Acryl Data (DataHub) | | Kind | Official | Official | | Repository or docs | [docs.atlan.com/product/capabilities/atlan-ai/references/mcp-tools](https://docs.atlan.com/product/capabilities/atlan-ai/references/mcp-tools) | [github.com/acryldata/mcp-server-datahub](https://github.com/acryldata/mcp-server-datahub) | | Version or commit | Hosted at mcp.atlan.com/mcp (unversioned); local atlan-mcp-server 0.3.3 (deprecated) | 0.8.0 | | Last release | 17 Feb 2026 | 21 Oct 2026 | | Transport | Remote (hosted) | Remote and local | | Auth | OAuth per user (calls run as the signed-in user) or Atlan API key as Bearer token (shared service identity); local server uses API key + base URL. | Stdio: DATAHUB_GMS_URL + DATAHUB_GMS_TOKEN (or ~/.datahubenv); HTTP: per-request DataHub personal access token as Authorization Bearer, no shared server token. | | Access | Configurable | Configurable | | Tools listed | 41 named in the evidence | 22 named in the evidence | | Tool definitions read at | [docs.atlan.com/product/capabilities/atlan-ai/references/mcp-tools](https://docs.atlan.com/product/capabilities/atlan-ai/references/mcp-tools) | [github.com/acryldata/mcp-server-datahub/blob/main/README.md](https://github.com/acryldata/mcp-server-datahub/blob/main/README.md) | | Context weight | 1,770 tokens for 42 tool definitions (42 with a description), measured from the docs page read | 567 tokens for 22 tool definitions (22 with a description), measured from the README read | | Status | Exists on 10 Nov 2026: hosted Atlan MCP (41 tools, page updated 27 Oct 2026); the local atlan-mcp-server in atlanhq/agent-toolkit is marked deprecated in its README. | Exists under this name on 10 Nov 2026; v0.8.0 released 21 Oct 2026 per GitHub releases and PyPI. | | Read | 10 Nov 2026 | 10 Nov 2026 | The same fourteen fields for both servers, read 10 Nov 2026. ## In one sentence each - [Atlan MCP Server](/servers/atlan-mcp/): Atlan's hosted per-tenant server runs as the signed-in user, scopes context by persona, offers an optional read-only mode and proposes writes for approval. - [DataHub MCP Server](/servers/datahub-mcp/): DataHub's server returns column-level lineage with paths and SQL transformations, plus glossary, domain and owner discovery, with per-user tokens in HTTP mode. ## The read facts, topic by topic | Topic | Atlan MCP Server | DataHub MCP Server | | --- | --- | --- | | Publisher and releases | Atlan hosts the MCP server itself at mcp.atlan.com/mcp, per tenant. The older local server in atlanhq/agent-toolkit is deprecated in favor of the hosted endpoint. The last GitHub release of the local server is v0.3.3, on 17 Feb 2026; the hosted server has no public version number. | mcp-server-datahub is published from the acryldata GitHub organization (DataHub's company) and is linked from DataHub's own docs at docs.datahub.com. The GitHub release page shows v0.8.0 as a published release (PyPI upload 21 Oct 2026). The package is on PyPI as mcp-server-datahub 0.8.0. | | Clients, auth and scoping | The MCP server serves the catalog as a context layer to AI clients (Claude, Cursor, ChatGPT, Gemini, Copilot) and automation platforms. OAuth makes each tool call run as the signed-in user, so tools run with their own permissions and actions are attributed to them. Guests and members are scoped to their persona; MCP narrows but never widens what a user's login permits. The tools page lists 41 tools across nine categories. | The server is built to serve DataHub context to AI agents: search, lineage, query history and entity metadata. In the Docker HTTP deployment each MCP client supplies its own DataHub token, so DataHub permissions and audit identity are preserved per user. An optional get_me tool (TOOLS_IS_USER_ENABLED=true) returns the authenticated user's profile and group memberships. | | Access and safety | An optional read-only mode removes every write, admin and lifecycle tool while keeping search and lineage; it is requested from Atlan support. Tools that change anything propose first and wait for approval before writing. Every tool call is logged as a structured event with tool name, client, tenant, request ID and status. Tools that execute SQL route it through a read-only validator that permits only SELECT-class queries. | Mutation tools are disabled by default and need TOOLS_IS_MUTATION_ENABLED=true, so the default is read-only. The HTTP entry point refuses to start if a shared server-wide token is configured, forcing per-user tokens. Tokens are accepted only in the Authorization header; query-string tokens are rejected. | | What it reaches | The docs list supported AI clients: Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, ChatGPT, Snowflake, Google ADK, n8n, Microsoft 365 Copilot, Glean and Databricks. Data-source coverage depends on Atlan connectors, not stated on this page. | The README says the server shows data in databases, lakes, warehouses and BI tools registered in DataHub; no connector list is on this page. | | Lineage and impact | The hosted Atlan MCP exposes a traverse_lineage tool for upstream/downstream lineage with deeper hops on request. The docs say the hosted Atlan MCP can traverse end-to-end lineage; the tool page does not state column-level or SQL-logic parsing for the lineage tool. None found: A statement on the MCP tools page that traverse_lineage returns column-level edges or transformation SQL. Query_assets runs read-only SQL against a connected warehouse (Snowflake, Redshift, BigQuery, Databricks), returning up to 100 rows. | DataHub's MCP server exposes a get_lineage tool covering datasets, columns and dashboards with hop control. The README says lineage is traceable at table and column level and that transformation logic is understood. A get_lineage_paths_between tool returns exact lineage paths between two assets or columns, including intermediate transformations and SQL query information. Get_dataset_queries fetches real SQL queries that reference a dataset or column. | | Definitions, ownership and rules | Tools cover glossaries, glossary categories and terms, domains and data products, custom metadata sets, and certification/descriptions. Add_atlan_tags applies classification tags with propagation to downstream lineage on by default. Resolve_metadata returns tenant-level tag and custom-metadata definitions reconciled live, including enum option values. The lineage tool returns owners, certificate status, tags and readme with each asset by default (local server docstring). | DataHub's MCP server lets agents discover domains, owners, tags and glossary terms. Optional mutation tools (off by default) add glossary terms, tags, owners, domains, descriptions and structured properties; this is metadata writing, and the docs do not say terms propagate along lineage through MCP. None found: Any statement in the DataHub MCP README that tags, terms or classifications propagate along lineage when read through the MCP tools. | | Freshness and upkeep | Metadata is retrieved in real time with no intermediate caching. | None found: Any freshness or ingestion-cadence statement in the DataHub MCP README (the server reads whatever DataHub has ingested). The README only describes caching of verified tokens for five minutes. | | Profile | Atlan's compliance posture for MCP is covered by HIPAA, GDPR and CCPA/CPRA programs, with certifications on the Atlan Trust Portal. | DataHub describes itself as an open-source context platform; the MCP server is Apache-2.0 licensed Python. | Each cell joins the facts read for that topic; the full list with sources is on each entry. Read 10 Nov 2026. ## Tool names, as published ### Atlan MCP Server `semantic_search`, `ask`, `fluent_search`, `fluent_search_schema`, `search_assets`, `get_assets`, `search_atlan_docs`, `resolve_metadata`, `get_groups`, `get_users`, `describe_asset_type`, `traverse_lineage`, `query_assets`, `update_assets`, `update_custom_metadata`, `add_atlan_tags`, `remove_atlan_tag`, `manage_announcements`, `create_custom_metadata_set`, `add_attributes_to_cm_set`, `remove_attributes_from_cm_set`, `remove_custom_metadata`, `delete_custom_metadata_set`, `manage_asset_lifecycle`, `create_glossaries`, `create_glossary_categories`, `create_glossary_terms`, `create_domains`, `create_data_products`, `create_dq_rules`, `update_dq_rules`, `schedule_dq_rules`, `delete_dq_rules`, `read_artifact`, `search_knowledge`, `list_knowledge`, `read_knowledge_file`, `create_knowledge_folder`, `save_knowledge_content`, `get_knowledge_upload_url`, `register_knowledge_file` ### DataHub MCP Server `search`, `get_lineage`, `get_dataset_queries`, `get_entities`, `get_aspect_history`, `list_schema_fields`, `get_lineage_paths_between`, `add_tags`, `remove_tags`, `add_terms`, `remove_terms`, `add_owners`, `remove_owners`, `set_domains`, `remove_domains`, `update_description`, `add_structured_properties`, `remove_structured_properties`, `get_me`, `search_documents`, `grep_documents`, `save_document` ## Read next - [Atlan MCP Server, the full entry](/servers/atlan-mcp/) - [DataHub MCP Server, the full entry](/servers/datahub-mcp/) - [Every side-by-side](/compare/) ## Sources - docs.atlan.com/product/capabilities/atlan-ai/references/mcp-tools: https://docs.atlan.com/product/capabilities/atlan-ai/references/mcp-tools (read 10 Nov 2026) - docs.atlan.com/product/capabilities/atlan-ai/how-tos/remote-mcp-overview: https://docs.atlan.com/product/capabilities/atlan-ai/how-tos/remote-mcp-overview (read 10 Nov 2026) - github.com/atlanhq/agent-toolkit/blob/main/modelcontextprotocol/server.py: https://github.com/atlanhq/agent-toolkit/blob/main/modelcontextprotocol/server.py (read 10 Nov 2026) - docs.atlan.com/product/capabilities/atlan-ai/references/mcp-security: https://docs.atlan.com/product/capabilities/atlan-ai/references/mcp-security (read 10 Nov 2026) - github.com/atlanhq/agent-toolkit/blob/main/modelcontextprotocol/README.md: https://github.com/atlanhq/agent-toolkit/blob/main/modelcontextprotocol/README.md (read 10 Nov 2026) - github.com/atlanhq/agent-toolkit/blob/main/CHANGELOG.md: https://github.com/atlanhq/agent-toolkit/blob/main/CHANGELOG.md (read 10 Nov 2026) - raw.githubusercontent.com/acryldata/mcp-server-datahub/main/README.md: https://raw.githubusercontent.com/acryldata/mcp-server-datahub/main/README.md (read 10 Nov 2026) - pypi.org/pypi/mcp-server-datahub/json: https://pypi.org/pypi/mcp-server-datahub/json (read 10 Nov 2026) - github.com/acryldata/mcp-server-datahub/releases/tag/v0.8.0: https://github.com/acryldata/mcp-server-datahub/releases/tag/v0.8.0 (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/compare/atlan-mcp-vs-datahub-mcp/ --- # Alation AI Agent SDK MCP server and Collibra MCP Server (chip), field by field: 12 and 49 tools named SIDE BY SIDE, read 10 Nov 2026 **The short answer.** Alation AI Agent SDK MCP server (Alation, remote and local) names 12 tools; Collibra MCP Server (chip) (Collibra, remote and local) names 49. Versions read on 10 Nov 2026: 1.0.0 and v0.0.67. The page puts the same fields in the same rows and draws no conclusion. | Field | Alation AI Agent SDK MCP server | Collibra MCP Server (chip) | | --- | --- | --- | | Data tool | [Alation](/mcp/alation/) | [Collibra](/mcp/collibra/) | | Publisher | Alation | Collibra | | Kind | Official | Official | | Repository or docs | [github.com/Alation/alation-ai-agent-sdk](https://github.com/Alation/alation-ai-agent-sdk) | [github.com/collibra/chip](https://github.com/collibra/chip) | | Version or commit | 1.0.0 | V0.0.67 | | Last release | 26 Oct 2026 | 27 Oct 2026 | | Transport | Remote and local | Remote and local | | Auth | OAuth client credentials of a service account (role-limited) or user refresh token; remote MCP at /ai/mcp needs client_id and client_secret (no dynamic client registration). | Collibra username/password: server-wide credentials over stdio (mcp.yaml or env vars), or client-supplied Basic Auth headers per request over HTTP; tools respect the connecting user's DGC scopes. | | Access | Not stated in the source read | Configurable | | Tools listed | 12 named in the evidence | 49 named in the evidence | | Tool definitions read at | [github.com/Alation/alation-ai-agent-sdk/tree/main/guides](https://github.com/Alation/alation-ai-agent-sdk/tree/main/guides) | [github.com/collibra/chip/blob/main/README.md](https://github.com/collibra/chip/blob/main/README.md) | | Context weight | 69 tokens for 4 tool definitions (3 with a description), measured from the docs page read | 2,914 tokens for 41 tool definitions (41 with a description), measured from the README read | | Status | Exists on 10 Nov 2026; alation-ai-agent-mcp 1.0.0 on PyPI and GitHub release v1.0.0 dated 26 Oct 2026. Access is read-only by documentation of removed write tools, not by a published read-only switch. | Exists on 10 Nov 2026 as collibra/chip; v0.0.67 released 27 Oct 2026 per GitHub releases. Collibra's product-docs page for it returned a Cloudflare block, so only the repo README was read. | | Read | 10 Nov 2026 | 10 Nov 2026 | The same fourteen fields for both servers, read 10 Nov 2026. ## In one sentence each - [Alation AI Agent SDK MCP server](/servers/alation-mcp/): Alation's server offers a lineage tool and a catalog search agent, and its remote form hides tools the current user cannot use. - [Collibra MCP Server (chip)](/servers/collibra-mcp/): Collibra's server covers lineage with transformation logic, business terms and column semantics, classification, stewardship edits and data quality, under the connecting user's permissions. ## The read facts, topic by topic | Topic | Alation AI Agent SDK MCP server | Collibra MCP Server (chip) | | --- | --- | --- | | Publisher and releases | Alation-ai-agent-mcp is published from the Alation GitHub organization and PyPI, version 1.0.0, released 26 Oct 2026. GitHub releases show v1.0.0 published 26 Oct 2026. The README calls 1.0.0 the first stable release of the SDK. | The repository lives under the collibra GitHub organization and is titled the Collibra MCP Server (Apache-2.0). GitHub releases show v0.0.67 published 27 Oct 2026, so the server is still at a 0.0.x version. The release timestamp for v0.0.67 is 27 Oct 2026. | | Clients, auth and scoping | The SDK and its MCP server let AI agents use Alation catalog metadata for curation, search and discovery, and data analyst agents. The remote MCP server lists tools dynamically, filtering out tools the current user cannot use. Service-account authentication gives the agent the data and features allowed by the role assigned to the service account. SQL tools need per-data-product data warehouse credentials configured once. | The server is described as giving AI agents access to Collibra Data Governance Center for asset discovery, glossary queries and asset details. Permission scoping follows the connecting Collibra user: tools needing extra DGC scopes are flagged, and a missing scope returns a permission error. Natural-language discovery tools use Collibra AI and consume Collibra Units. With the HTTP transport, MCP clients are advised to supply their own Basic Auth headers per request; with stdio, one server-wide credential is used. | | Access and safety | Metadata-writing tools update_catalog_asset_metadata and check_job_status were removed from the SDK (Jan 2026), leaving curation to the REST API. The MCP server supports an explicit enabled-tools list (CLI argument or ALATION_ENABLED_TOOLS) to limit the tools exposed. The remote MCP server does not support dynamic client registration; clients must supply a client_id and client_secret. | Data quality tools (18 tools, several that write or delete) are off by default and gated behind a --data-quality flag. Specific tools can be enabled or disabled through flags, environment variables or mcp.yaml; by default all are enabled. Destructive DQ tools use a confirm checkpoint (confirm=false by default is read-only). HTTP mode binds to localhost only, and credentials should be kept in environment variables. | | What it reaches | None found: A list of warehouses, transformation or BI tools in the Alation agent SDK README (coverage follows the Alation catalog). | None found: A list of warehouses, BI tools or transformation tools supported by the Collibra MCP README (coverage depends on what is catalogued in the Collibra instance). | | Lineage and impact | Alation's lineage tool traverses upstream or downstream from a starting object and supports column-level lineage; it is a BETA feature that Alation support must enable on the instance. The lineage tool takes max_depth (default 10), allowed_otypes, design/run-time filters and schema exclusions. None found: Documentation that the lineage tool returns transformation SQL/logic or BI-field/metric nodes; the tool page lists graph nodes filtered by object type only. | The Collibra MCP server has separate tools for upstream and downstream technical lineage, lineage entities, and transformations. Get_lineage_transformation returns the details and logic of a specific data transformation. Lineage tools search entities and transformations in the technical lineage graph. Get_measure_data traces a measure back to its underlying physical columns and tables. None found: A README statement that the lineage tools return column-level edges or BI-field lineage explicitly; the tool list names entities and transformations only. | | Definitions, ownership and rules | The lineage docs show combining lineage with the Aggregated Context tool to return stewards or an Owner custom field, and the data dictionary tool for column fields. Get_custom_fields_definitions and the data dictionary tools are listed as supported tools for custom metadata. The lineage guide describes updating a PII flag across downstream columns found via lineage, done by the LLM through tools (not automatic propagation). | Get_business_term_data traces a business term back to its connected physical data assets. Get_column_semantics returns data attributes, measures and business assets connected to a column; get_table_semantics returns the table's semantic layer. Edit_asset can set attributes such as Definition and assign users or groups to resource roles such as Steward and Owner. Classification tools associate data classes with assets (add_data_classification_match, search_data_classification_match). | | Freshness and upkeep | None found: A freshness or collection-cadence statement in the Alation agent SDK README or lineage guide. | None found: Any statement on collection cadence or freshness in the Collibra MCP README (the server queries a live Collibra instance). | | Profile | The remote MCP server is served from the customer's own Alation instance at /ai/mcp. | The server is a Go binary distributed as prebuilt releases for Linux, macOS and Windows. | Each cell joins the facts read for that topic; the full list with sources is on each entry. Read 10 Nov 2026. ## Tool names, as published ### Alation AI Agent SDK MCP server `alation_context`, `bulk_retrieval`, `data_quality_tool`, `get_custom_fields_definitions`, `get_data_products`, `get_data_sources_tool`, `get_signature_creation_instructions`, `lineage`, `catalog_context_search_agent`, `query_flow_agent`, `sql_query_agent`, `custom_agent` ### Collibra MCP Server (chip) `discover_business_glossary`, `discover_data_assets`, `get_assessment`, `get_asset_details`, `get_business_term_data`, `get_column_semantics`, `get_data_quality_rule`, `get_data_quality_rule_results`, `validate_data_quality_rule`, `list_data_quality_rule_templates`, `get_data_quality_rule_template`, `find_data_quality_rules`, `generate_data_quality_rule_sql`, `get_lineage_downstream`, `get_lineage_entity`, `get_lineage_transformation`, `get_lineage_upstream`, `get_measure_data`, `get_table_semantics`, `get_user_id_by_name`, `list_asset_types`, `list_data_contract`, `prepare_create_asset`, `pull_data_contract_manifest`, `search_asset_keyword`, `search_catalog_columns`, `search_data_class`, `search_data_classification_match`, `search_lineage_entities`, `search_lineage_transformations`, `add_data_classification_match`, `create_assessment`, `create_asset`, `create_data_quality_job`, `create_data_quality_rule`, `deploy_data_quality_rule_template`, `dq_cancel_job_run`, `dq_delete_job`, `dq_delete_job_run`, `dq_update_job`, `dq_get_job`, `dq_get_job_run`, `dq_search_jobs`, `dq_search_job_runs`, `edit_assessment`, `edit_asset`, `init_data_contract`, `push_data_contract_manifest`, `remove_data_classification_match` ## Read next - [Alation AI Agent SDK MCP server, the full entry](/servers/alation-mcp/) - [Collibra MCP Server (chip), the full entry](/servers/collibra-mcp/) - [Every side-by-side](/compare/) ## Sources - github.com/Alation/alation-ai-agent-sdk/blob/main/guides/tools/lineage.md: https://github.com/Alation/alation-ai-agent-sdk/blob/main/guides/tools/lineage.md (read 10 Nov 2026) - github.com/Alation/alation-ai-agent-sdk/blob/main/README.md: https://github.com/Alation/alation-ai-agent-sdk/blob/main/README.md (read 10 Nov 2026) - github.com/Alation/alation-ai-agent-sdk/blob/main/guides/authentication.md: https://github.com/Alation/alation-ai-agent-sdk/blob/main/guides/authentication.md (read 10 Nov 2026) - pypi.org/pypi/alation-ai-agent-mcp/json: https://pypi.org/pypi/alation-ai-agent-mcp/json (read 10 Nov 2026) - github.com/Alation/alation-ai-agent-sdk/releases: https://github.com/Alation/alation-ai-agent-sdk/releases (read 10 Nov 2026) - github.com/collibra/chip/blob/main/README.md: https://github.com/collibra/chip/blob/main/README.md (read 10 Nov 2026) - github.com/collibra/chip/releases: https://github.com/collibra/chip/releases (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/compare/alation-mcp-vs-collibra-mcp/ --- # DataHub MCP Server and OpenMetadata MCP Server, field by field: 22 and 15 tools named SIDE BY SIDE, read 10 Nov 2026 **The short answer.** DataHub MCP Server (Acryl Data, remote and local) names 22 tools; OpenMetadata MCP Server (OpenMetadata, remote (hosted)) names 15. Versions read on 10 Nov 2026: 0.8.0 and 2.0.7. The page puts the same fields in the same rows and draws no conclusion. | Field | DataHub MCP Server | OpenMetadata MCP Server | | --- | --- | --- | | Data tool | [DataHub](/mcp/datahub/) | [OpenMetadata](/mcp/openmetadata/) | | Publisher | Acryl Data (DataHub) | OpenMetadata | | Kind | Official | Official | | Repository or docs | [github.com/acryldata/mcp-server-datahub](https://github.com/acryldata/mcp-server-datahub) | [docs.open-metadata.org/v2.0.x/how-to-guides/mcp.md](https://docs.open-metadata.org/v2.0.x/how-to-guides/mcp.md) | | Version or commit | 0.8.0 | 2.0.7 (bundled with platform release 2.0.7-release; 2.1.0-rc1 also listed) | | Last release | 21 Oct 2026 | 27 Oct 2026 | | Transport | Remote and local | Remote (hosted) | | Auth | Stdio: DATAHUB_GMS_URL + DATAHUB_GMS_TOKEN (or ~/.datahubenv); HTTP: per-request DataHub personal access token as Authorization Bearer, no shared server token. | OAuth 2.0 via existing SSO (per user), personal access token, or Bot JWT as Authorization Bearer. | | Access | Configurable | Read and write | | Tools listed | 22 named in the evidence | 15 named in the evidence | | Tool definitions read at | [github.com/acryldata/mcp-server-datahub/blob/main/README.md](https://github.com/acryldata/mcp-server-datahub/blob/main/README.md) | [docs.open-metadata.org/v2.0.x/how-to-guides/mcp/reference.md](https://docs.open-metadata.org/v2.0.x/how-to-guides/mcp/reference.md) | | Context weight | 567 tokens for 22 tool definitions (22 with a description), measured from the README read | 1,249 tokens for 15 tool definitions (15 with a description), measured from the docs page read | | Status | Exists under this name on 10 Nov 2026; v0.8.0 released 21 Oct 2026 per GitHub releases and PyPI. | Exists on 10 Nov 2026; built into the OpenMetadata server (not a separate package). Tool access is governed by the caller's role and policies; no read-only switch is documented on the pages read. | | Read | 10 Nov 2026 | 10 Nov 2026 | The same fourteen fields for both servers, read 10 Nov 2026. ## In one sentence each - [DataHub MCP Server](/servers/datahub-mcp/): DataHub's server returns column-level lineage with paths and SQL transformations, plus glossary, domain and owner discovery, with per-user tokens in HTTP mode. - [OpenMetadata MCP Server](/servers/openmetadata-mcp/): OpenMetadata's built-in server offers per-user OAuth, persona-scoped context, table and optional column lineage, and write tools for lineage, entities and tests. ## The read facts, topic by topic | Topic | DataHub MCP Server | OpenMetadata MCP Server | | --- | --- | --- | | Publisher and releases | mcp-server-datahub is published from the acryldata GitHub organization (DataHub's company) and is linked from DataHub's own docs at docs.datahub.com. The GitHub release page shows v0.8.0 as a published release (PyPI upload 21 Oct 2026). The package is on PyPI as mcp-server-datahub 0.8.0. | The MCP server is built into OpenMetadata and installed and enabled by default. OpenMetadata releases 2.0.7 on 27 Oct 2026 (GitHub release page); the MCP server version follows the platform release. The release timestamp is 27 Oct 2026. | | Clients, auth and scoping | The server is built to serve DataHub context to AI agents: search, lineage, query history and entity metadata. In the Docker HTTP deployment each MCP client supplies its own DataHub token, so DataHub permissions and audit identity are preserved per user. An optional get_me tool (TOOLS_IS_USER_ENABLED=true) returns the authenticated user's profile and group memberships. | Get_user_context returns the authenticated caller's identity, roles, teams, domains and active persona, with no parameter to look up another user. A personal access token gives the agent the same role and access policy as the user. OAuth 2.0 runs tool calls with the user's existing SSO login (Google, Azure, Okta, Auth0, LDAP, SAML). | | Access and safety | Mutation tools are disabled by default and need TOOLS_IS_MUTATION_ENABLED=true, so the default is read-only. The HTTP entry point refuses to start if a shared server-wide token is configured, forcing per-user tokens. Tokens are accepted only in the Authorization header; query-string tokens are rejected. | Every tool is labeled Read or Write in the reference. Unattended agents should use a Bot service account with a role-based access policy, kept separate from personal logins and auditable. The transport is Streamable HTTP (MCP spec 26 Mar 2025). | | What it reaches | The README says the server shows data in databases, lakes, warehouses and BI tools registered in DataHub; no connector list is on this page. | The reference lists asset types: database, dashboard, pipeline, topic, mlmodel, container, searchIndex, storedProcedure and API services; no named vendor list is on this page. | | Lineage and impact | DataHub's MCP server exposes a get_lineage tool covering datasets, columns and dashboards with hop control. The README says lineage is traceable at table and column level and that transformation logic is understood. A get_lineage_paths_between tool returns exact lineage paths between two assets or columns, including intermediate transformations and SQL query information. Get_dataset_queries fetches real SQL queries that reference a dataset or column. | Get_entity_lineage returns upstream and downstream lineage for any entity, with an includeColumnLineage option (default false, table level only), up to 10 hops. Create_lineage lets an agent write lineage edges, and root_cause_analysis is a lineage-and-impact tool. None found: Documentation that the MCP lineage tools return transformation SQL or logic. | | Definitions, ownership and rules | DataHub's MCP server lets agents discover domains, owners, tags and glossary terms. Optional mutation tools (off by default) add glossary terms, tags, owners, domains, descriptions and structured properties; this is metadata writing, and the docs do not say terms propagate along lineage through MCP. None found: Any statement in the DataHub MCP README that tags, terms or classifications propagate along lineage when read through the MCP tools. | Search_metadata filters on owners, tags, tiers and glossary terms and entity types include glossary, glossaryTerm, metric, domain and dataProduct. Find_context and get_persona_context serve shared AI context documents curated per persona. Patch_entity and create_entity write metadata (create never modifies an existing entity; patch handles edits). | | Freshness and upkeep | None found: Any freshness or ingestion-cadence statement in the DataHub MCP README (the server reads whatever DataHub has ingested). The README only describes caching of verified tokens for five minutes. | None found: Freshness or collection-cadence statement on the OpenMetadata MCP pages. | | Profile | DataHub describes itself as an open-source context platform; the MCP server is Apache-2.0 licensed Python. | OpenMetadata's MCP server is described as embedded in an open knowledge graph of metadata, with supported SSO providers including Google, Azure, Okta, Auth0, LDAP and SAML. | Each cell joins the facts read for that topic; the full list with sources is on each entry. Read 10 Nov 2026. ## Tool names, as published ### DataHub MCP Server `search`, `get_lineage`, `get_dataset_queries`, `get_entities`, `get_aspect_history`, `list_schema_fields`, `get_lineage_paths_between`, `add_tags`, `remove_tags`, `add_terms`, `remove_terms`, `add_owners`, `remove_owners`, `set_domains`, `remove_domains`, `update_description`, `add_structured_properties`, `remove_structured_properties`, `get_me`, `search_documents`, `grep_documents`, `save_document` ### OpenMetadata MCP Server `search_metadata`, `semantic_search`, `company_context`, `get_entity_details`, `get_user_context`, `get_persona_context`, `find_context`, `get_entity_lineage`, `create_lineage`, `root_cause_analysis`, `create_entity`, `describe_entity_type`, `patch_entity`, `get_test_definitions`, `create_test_case` ## Read next - [DataHub MCP Server, the full entry](/servers/datahub-mcp/) - [OpenMetadata MCP Server, the full entry](/servers/openmetadata-mcp/) - [Every side-by-side](/compare/) ## Sources - raw.githubusercontent.com/acryldata/mcp-server-datahub/main/README.md: https://raw.githubusercontent.com/acryldata/mcp-server-datahub/main/README.md (read 10 Nov 2026) - pypi.org/pypi/mcp-server-datahub/json: https://pypi.org/pypi/mcp-server-datahub/json (read 10 Nov 2026) - github.com/acryldata/mcp-server-datahub/releases/tag/v0.8.0: https://github.com/acryldata/mcp-server-datahub/releases/tag/v0.8.0 (read 10 Nov 2026) - docs.open-metadata.org/v2.0.x/how-to-guides/mcp/reference.md: https://docs.open-metadata.org/v2.0.x/how-to-guides/mcp/reference.md (read 10 Nov 2026) - docs.open-metadata.org/v2.0.x/how-to-guides/mcp.md: https://docs.open-metadata.org/v2.0.x/how-to-guides/mcp.md (read 10 Nov 2026) - github.com/open-metadata/OpenMetadata/releases/tag/2.0.7-release: https://github.com/open-metadata/OpenMetadata/releases/tag/2.0.7-release (read 10 Nov 2026) - docs.open-metadata.org/v2.0.x/how-to-guides/mcp/connect.md: https://docs.open-metadata.org/v2.0.x/how-to-guides/mcp/connect.md (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/compare/datahub-mcp-vs-openmetadata-mcp/ --- # Secoda MCP and Select Star MCP Server, field by field: 8 and 7 tools named SIDE BY SIDE, read 10 Nov 2026 **The short answer.** Secoda MCP (Secoda, remote and local) names 8 tools; Select Star MCP Server (Select Star, remote (hosted)) names 7. Versions read on 10 Nov 2026: no version number published and no version number published. The page puts the same fields in the same rows and draws no conclusion. | Field | Secoda MCP | Select Star MCP Server | | --- | --- | --- | | Data tool | [Secoda](/mcp/secoda/) | [Select Star](/mcp/select-star/) | | Publisher | Secoda | Select Star (Snowflake agreed to acquire it, announced 24 Nov 2025) | | Kind | Official | Official | | Repository or docs | [github.com/secoda/secoda-mcp](https://github.com/secoda/secoda-mcp) | [docs.selectstar.com/features/mcp-server](https://docs.selectstar.com/features/mcp-server) | | Version or commit | No releases; main branch last commit 26 Jan 2026 | Hosted, no version published | | Last release | none found on 10 Nov 2026 | none found on 10 Nov 2026 | | Transport | Remote and local | Remote (hosted) | | Auth | Secoda API token (API_TOKEN env var) for the local FastMCP wrapper; docs also describe a Secoda-hosted MCP endpoint authenticated against workspace and AI permissions. | Select Star API token sent as Authorization: Bearer header (mcp-remote proxy used for Claude Desktop). | | Access | Not stated in the source read | Read only | | Tools listed | 8 named in the evidence | 7 named in the evidence | | Tool definitions read at | [github.com/secoda/secoda-mcp/blob/main/server.py](https://github.com/secoda/secoda-mcp/blob/main/server.py) | [docs.selectstar.com/features/mcp-server](https://docs.selectstar.com/features/mcp-server) | | Context weight | 63 tokens for 6 tool definitions (6 with a description), measured from the source code read | 98 tokens for 7 tool definitions (7 with a description), measured from the docs page read | | Status | Exists on 10 Nov 2026: open-source wrapper with 10 commits and no GitHub releases, plus a Secoda-hosted MCP described in docs.secoda.co (eight tools listed there; the wrapper code implements run_sql, search_data_assets, search_documentation, retrieve_entity, entity_lineage and glossary). | Exists on 10 Nov 2026 at mcp.production.selectstar.com/mcp. docs.selectstar.com did not resolve from the runner today, so the docs page was read from a web.archive.org snapshot dated 14 Jun 2026; Snowflake announced plans to acquire Select Star on 24 Nov 2025. | | Read | 10 Nov 2026 | 10 Nov 2026 | The same fourteen fields for both servers, read 10 Nov 2026. ## In one sentence each - [Secoda MCP](/servers/secoda-mcp/): Secoda's server combines catalog search, glossary, lineage and SQL over a workspace, as a hosted endpoint and an open-source wrapper. - [Select Star MCP Server](/servers/select-star-mcp/): Select Star's hosted, read-only server exposes catalog search, lineage, query history and semantic models with token setup. ## The read facts, topic by topic | Topic | Secoda MCP | Select Star MCP Server | | --- | --- | --- | | Publisher and releases | Secoda-mcp is under the secoda GitHub organization and documented in Secoda's own docs; it is MIT licensed Python/FastMCP. GitHub shows no releases for secoda-mcp, so no version number exists. The latest commit on main is a merge of pull request 4 dated 26 Jan 2026. | The server is hosted by Select Star at mcp.production.selectstar.com/mcp and documented on Select Star's own docs site. No version number or release date is published for the MCP server; the docs page was last updated 7 months before the 14 Jun 2026 archive snapshot read. | | Clients, auth and scoping | The Secoda-hosted MCP server lets AI tools search, query SQL, read lineage and glossary terms. Authentication verifies the user's workspace and AI permissions, and operations respect governance and access controls. The open-source wrapper authenticates with a single API_TOKEN, i.e. one shared token per server process. | The MCP server is meant to bring catalog metadata and lineage to Cursor, Claude Desktop and VS Code GitHub Copilot. Authentication is a Select Star API token sent as a Bearer header; no per-user OAuth is described, and the docs do not say results are scoped to the asking user. None found: A statement that Select Star MCP results are limited to the asking user's own permissions. | | Access and safety | The tools shown are search, SQL execution, entity retrieval, lineage and glossary; run_sql can execute SQL, and the docs do not describe a read-only mode. Admin must enable MCP and the user needs the right Secoda permissions. | Seven tools are listed, all search/get/traverse operations; no write tools appear on the page. Troubleshooting says the token must be formatted with a Bearer prefix. | | What it reaches | None found: A list of supported warehouses or BI tools on the Secoda MCP page. | Select Star says its universal metadata model connects more than 25 integrations (founder post on the Snowflake deal); the MCP page names no systems. | | Lineage and impact | The entity_lineage tool explores upstream and downstream dependencies for an entity. In the open-source wrapper entity_lineage takes only an entity_id and no depth or column parameter. None found: Documentation of column-level lineage, transformation SQL, BI-field or metric lineage in the Secoda MCP docs. The run_sql tool executes SQL against connected warehouses. | The Select Star MCP server has a traverse_lineage tool for lineage relationships and dependencies. Get_asset_queries returns warehouse queries connected to tables or columns. None found: Documentation on the MCP page that traverse_lineage is column-level or returns transformation SQL, BI-field or metric lineage. | | Definitions, ownership and rules | A glossary tool returns business terms and definitions. None found: Ownership, classification or governance-rule propagation along lineage in the Secoda MCP docs. | The product page says the MCP server returns descriptions, tags, owners and popularity metrics for agents. It exposes the semantic layer and business glossary, with standardized metrics, dimensions and business terms. Two tools search semantic models and return their full details including YAML. | | Freshness and upkeep | None found: Freshness statement in the Secoda MCP docs. | None found: Any freshness or collection-cadence statement on the Select Star MCP docs page. | | Profile | Secoda supports EU, APAC and self-hosted instances by changing the API URL. | Snowflake announced plans to acquire the Select Star platform (post dated 24 Nov 2025). | Each cell joins the facts read for that topic; the full list with sources is on each entry. Read 10 Nov 2026. ## Tool names, as published ### Secoda MCP `search_data_assets`, `search`, `run_sql`, `retrieve_entity`, `entity_lineage`, `glossary`, `get_secoda_docs`, `chart` ### Select Star MCP Server `search_metadata`, `get_asset_details`, `get_table_info`, `traverse_lineage`, `get_asset_queries`, `search_semantic_models`, `get_semantic_model_details` ## Read next - [Secoda MCP, the full entry](/servers/secoda-mcp/) - [Select Star MCP Server, the full entry](/servers/select-star-mcp/) - [Every side-by-side](/compare/) ## Sources - docs.secoda.co/features/ai-assistant/secoda-mcp-server.md: https://docs.secoda.co/features/ai-assistant/secoda-mcp-server.md (read 10 Nov 2026) - github.com/secoda/secoda-mcp/blob/main/server.py: https://github.com/secoda/secoda-mcp/blob/main/server.py (read 10 Nov 2026) - github.com/secoda/secoda-mcp/releases: https://github.com/secoda/secoda-mcp/releases (read 10 Nov 2026) - api.github.com/repos/secoda/secoda-mcp/commits?per_page=1: https://api.github.com/repos/secoda/secoda-mcp/commits?per_page=1 (read 10 Nov 2026) - docs.selectstar.com/features/mcp-server: https://docs.selectstar.com/features/mcp-server (read 10 Nov 2026) - selectstar.com/product/mcp-for-data: https://selectstar.com/product/mcp-for-data (read 10 Nov 2026) - selectstar.com/resources/snowflake-acquire-select-star: https://selectstar.com/resources/snowflake-acquire-select-star (read 10 Nov 2026) - web.archive.org/web/20260614045618/https://docs.selectstar.com/features/mcp-server: https://web.archive.org/web/20260614045618/https://docs.selectstar.com/features/mcp-server (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/compare/secoda-mcp-vs-select-star-mcp/ --- # Monte Carlo MCP Server and SYNQ MCP (Coalesce Quality), field by field: 15 and 134 tools named SIDE BY SIDE, read 10 Nov 2026 **The short answer.** Monte Carlo MCP Server (Monte Carlo, remote (hosted)) names 15 tools; SYNQ MCP (Coalesce Quality) (SYNQ, remote (hosted)) names 134. Versions read on 10 Nov 2026: no version number published and no version number published. The page puts the same fields in the same rows and draws no conclusion. | Field | Monte Carlo MCP Server | SYNQ MCP (Coalesce Quality) | | --- | --- | --- | | Data tool | [Monte Carlo](/mcp/monte-carlo/) | [SYNQ](/mcp/synq/) | | Publisher | Monte Carlo | SYNQ (docs titled Coalesce Quality) | | Kind | Official | Official | | Repository or docs | [docs.getmontecarlo.com/docs/mcp-server](https://docs.getmontecarlo.com/docs/mcp-server) | [docs.synq.io/scout/mcp.md](https://docs.synq.io/scout/mcp.md) | | Version or commit | Hosted at mcp.getmontecarlo.com/mcp, no version published | Hosted, no version published | | Last release | none found on 10 Nov 2026 | none found on 10 Nov 2026 | | Transport | Remote (hosted) | Remote (hosted) | | Auth | OAuth 2.1 with Dynamic Client Registration (user identity, Editor role or above), or MCP-server-type API key id:secret as Bearer token via mcp-remote. | OAuth 2 with PKCE per user with selectable permission tier (read, write, governance, deploy); legacy workspace token (st-...) as Bearer. | | Access | Configurable | Configurable | | Tools listed | 15 named in the evidence | 134 named in the evidence | | Tool definitions read at | none found on 10 Nov 2026 | [docs.synq.io/scout/mcp.md](https://docs.synq.io/scout/mcp.md) | | Context weight | Not measured: no published tool-definition text was read on 10 Nov 2026 | 4,550 tokens for 134 tool definitions (134 with a description), measured from the docs page read | | Status | Exists on 10 Nov 2026 as a hosted first-party server; tool names are partial (those named in the docs), full tool definitions were not published as text, so no schema file was saved. | Exists on 10 Nov 2026; docs.synq.io now titles itself Coalesce Quality Documentation and its MCP page calls the product Coalesce Quality; regional hosted endpoints EU/US/AU. | | Read | 10 Nov 2026 | 10 Nov 2026 | The same fourteen fields for both servers, read 10 Nov 2026. ## In one sentence each - [Monte Carlo MCP Server](/servers/monte-carlo-mcp/): Monte Carlo's hosted observability server uses per-user OAuth or scoped keys, a per-request read-only header and tool allow-lists. - [SYNQ MCP (Coalesce Quality)](/servers/synq-mcp/): SYNQ's hosted observability and lineage server uses per-user OAuth with tiered consent scopes, and tools hide when a scope is not granted. ## The read facts, topic by topic | Topic | Monte Carlo MCP Server | SYNQ MCP (Coalesce Quality) | | --- | --- | --- | | Publisher and releases | The server is first-party, run by Monte Carlo at mcp.getmontecarlo.com/mcp, not distributed as a package. No version number or public release date is published for the hosted server; the docs page carries an updatedAt of 2 Nov 2026. | The server is hosted by SYNQ at regional URLs (EU mcp.synq.io, US, AU) and needs no workspace setup; no version or release date is published. None found: A GitHub repository, package version or release date for the SYNQ MCP server on the docs page. | | Clients, auth and scoping | Monte Carlo says connected agents can investigate alerts, explore assets and lineage, and create and manage monitors. OAuth lets any HTTP-transport MCP client connect with no API keys, and the connector is listed in the Claude Connectors Directory. An Editor role or above is required to use the server. An agent_observability toolset adds get_agent_metadata, get_agent_conversation and get_agent_trace for inspecting AI agents. | OAuth 2 with PKCE means actions are recorded as the real user with a via MCP badge. The set of tools an assistant sees depends on the permissions the user grants at consent, and the consent screen offers only what the user's role allows. The iam tool reports the caller's workspace, granted scopes and user. | | Access and safety | An x-mcp-readonly header hides write tools such as update_alert and create_or_update_* monitors and blocks calls to them. Per-request x-mcp-tools allow-lists narrow tools and are enforced on call, not only on listing. API-key access uses MCP-server-specific scoped keys; standard API keys do not work. The MCP server can be restricted by IP using Network Access Control Lists. External security audits and penetration tests are done at least annually, covering MCP surfaces. | Four permission tiers (read-only, write, governance, deploy) each grant only the scopes asked for; read-only makes no changes. Write tools do not appear in the client if the user granted only read-only access. Admins can switch off OAuth access to the API for third-party clients, including MCP, with the Disable user API access setting; legacy static workspace tokens (st-...) are also accepted. | | What it reaches | None found: A list of warehouses and BI systems on the MCP pages (the server fronts the Monte Carlo account's monitored assets). | The docs name dbt, SQLMesh, Coalesce, Airflow and warehouse query history as sources the tools draw on. | | Lineage and impact | Monte Carlo's MCP server includes a get_asset_lineage tool, used to see how data flows into a table. None found: Documentation that get_asset_lineage returns column-level lineage, transformation SQL, BI fields or metrics; the MCP pages name only asset lineage. | SYNQ (docs now titled Coalesce Quality) serves the same lineage graph as its lineage view, including column-level lineage where the warehouse supports it. Lineage tools include get_lineage (up/down/both with depth), upstream and downstream dependencies, common upstream, and upstream sources such as dbt sources. Get_code returns the latest source code defining an entity, including SQL definitions and dbt models. Impact tools list the entities and systems affected by an entity or an issue. | | Definitions, ownership and rules | None found: Glossary terms, definitions or classification tools in the Monte Carlo MCP server docs; the tools are alerts, monitors, assets, lineage and data operations. Governance-like context comes from use cases and critical-table groupings (get_use_cases, get_use_case_tables) in the extended toolset. | Knowledge tools read entity descriptions and knowledge pages pinned to entities, including owners and certified pages. The Governance permission tier lets the assistant manage data products, domains, owners, alert routing and check categories. List_annotations and batch_entity_annotations read annotations carried by entities. | | Freshness and upkeep | None found: A freshness statement on the MCP pages (the server reads live alert and monitor state from the Monte Carlo account). | Tools return execution history and latest executions for monitors, dbt models and Airflow tasks; join patterns use the last 30 days of warehouse query history. | | Profile | Monte Carlo also publishes an Agent Toolkit (mc-agent-toolkit) of skills for monitoring and code-change validation, separate from the server. | SYNQ's documentation is titled Coalesce Quality Documentation; the relationship between the names is not explained on the pages read. | Each cell joins the facts read for that topic; the full list with sources is on each entry. Read 10 Nov 2026. ## Tool names, as published ### Monte Carlo MCP Server `search`, `get_table`, `get_asset_lineage`, `get_alerts`, `alert_assessment`, `update_alert`, `get_monitors`, `create_or_update_table_monitor`, `get_use_cases`, `get_use_case_tables`, `get_use_case_table_summary`, `get_agent_metadata`, `get_agent_conversation`, `get_agent_trace`, `create_agent_metric_monitor` ### SYNQ MCP (Coalesce Quality) `add_comment`, `apply_service_account_changes`, `batch_entity_annotations`, `batch_entity_details`, `batch_executions`, `batch_urls`, `browse_folders`, `create_incident`, `create_monitors`, `create_sql_tests`, `create_table_stats_monitor`, `decide_recommendation`, `delete_alert`, `delete_check_categorisation_rule`, `delete_data_product`, `delete_deployment_rule`, `delete_domain`, `delete_monitors`, `delete_owner`, `delete_saved_view`, `delete_sql_test_deployment_rule`, `delete_sql_tests`, `deploy_deployment_rule`, `deploy_sql_test_deployment_rule`, `describe_check_categorisation`, `describe_check_types`, `describe_resolver_ql`, `exclude_from_monitoring`, `execute_monitor`, `folder_of`, `get_alert`, `get_code`, `get_commit_diff`, `get_common_upstream_dependencies`, `get_coverage_summary`, `get_data_product`, `get_database_change_details`, `get_database_coordinates_from_entity`, `get_deployment_rule`, `get_domain`, `get_downstream_dependencies`, `get_entity_access`, `get_entity_details`, `get_entity_from_database_coordinates`, `get_entity_impact`, `get_entity_knowledge`, `get_entity_metrics`, `get_entity_overlays`, `get_immediate_downstream_dependencies`, `get_immediate_upstream_dependencies`, `get_incident_details`, `get_issue`, `get_issue_comments`, `get_issue_details`, `get_issue_impact`, `get_join_patterns`, `get_knowledge_pages`, `get_latest_executions`, `get_lineage`, `get_monitor`, `get_monitor_predictions`, `get_owner`, `get_query_activity`, `get_saved_view`, `get_schema`, `get_sql_test_deployment_rule`, `get_upstream_dependencies`, `get_upstream_sources`, `iam`, `list_access_signals`, `list_alerts`, `list_annotations`, `list_categorised_checks`, `list_check_categories`, `list_check_categorisation_rules`, `list_checks`, `list_commits`, `list_data_product_members`, `list_data_products`, `list_database_changes`, `list_deployment_rules`, `list_domain_members`, `list_domains`, `list_entity_coverage`, `list_executions`, `list_folder_members`, `list_history`, `list_integration_health`, `list_knowledge_pages`, `list_monitor_segments`, `list_monitoring_exclusions`, `list_monitors`, `list_open_incidents`, `list_open_issues`, `list_owners`, `list_ownerships`, `list_platform_users`, `list_recommendations`, `list_saved_views`, `list_service_account_suggestions`, `list_service_accounts`, `list_sql_test_deployment_rules`, `list_sql_tests`, `list_table_stats_monitors`, `list_users`, `plan_service_account_changes`, `preview_check_categorisation_rule`, `preview_deployment_rule`, `preview_sql_test_deployment_rule`, `profile_columns`, `resolve_query`, `run_sql_test_again`, `sample_column_values`, `sample_entities`, `save_alert`, `save_check_categorisation_rule`, `save_data_product`, `save_domain`, `save_owner`, `save_saved_view`, `search_entities`, `send_test_alert`, `set_alert_enabled`, `set_check_category`, `set_issue_status`, `set_monitor_segment_labels`, `set_saved_view_pin`, `submit_feature_request`, `summarise_executions`, `unexclude_from_monitoring`, `update_incident`, `update_monitors`, `update_sql_tests`, `update_table_stats_monitor` ## Read next - [Monte Carlo MCP Server, the full entry](/servers/monte-carlo-mcp/) - [SYNQ MCP (Coalesce Quality), the full entry](/servers/synq-mcp/) - [Every side-by-side](/compare/) ## Sources - docs.getmontecarlo.com/docs/mcp-server: https://docs.getmontecarlo.com/docs/mcp-server (read 10 Nov 2026) - docs.getmontecarlo.com/docs/mcp-server-reference-and-operations: https://docs.getmontecarlo.com/docs/mcp-server-reference-and-operations (read 10 Nov 2026) - docs.getmontecarlo.com/docs/ai-security: https://docs.getmontecarlo.com/docs/ai-security (read 10 Nov 2026) - docs.synq.io/scout/mcp.md: https://docs.synq.io/scout/mcp.md (read 10 Nov 2026) - docs.synq.io/llms.txt: https://docs.synq.io/llms.txt (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/compare/monte-carlo-mcp-vs-synq-mcp/ --- # Monte Carlo MCP Server and Sifflet MCP Server, field by field: 15 and 10 tools named SIDE BY SIDE, read 10 Nov 2026 **The short answer.** Monte Carlo MCP Server (Monte Carlo, remote (hosted)) names 15 tools; Sifflet MCP Server (Sifflet, local (stdio)) names 10. Versions read on 10 Nov 2026: no version number published and 0.1.7. The page puts the same fields in the same rows and draws no conclusion. | Field | Monte Carlo MCP Server | Sifflet MCP Server | | --- | --- | --- | | Data tool | [Monte Carlo](/mcp/monte-carlo/) | [Sifflet](/mcp/sifflet/) | | Publisher | Monte Carlo | Sifflet | | Kind | Official | Official | | Repository or docs | [docs.getmontecarlo.com/docs/mcp-server](https://docs.getmontecarlo.com/docs/mcp-server) | [github.com/siffletdata/sifflet-mcp](https://github.com/siffletdata/sifflet-mcp) | | Version or commit | Hosted at mcp.getmontecarlo.com/mcp, no version published | 0.1.7 (PyPI); main at commit 1f390ca | | Last release | none found on 10 Nov 2026 | 3 Aug 2026 | | Transport | Remote (hosted) | Local (stdio) | | Auth | OAuth 2.1 with Dynamic Client Registration (user identity, Editor role or above), or MCP-server-type API key id:secret as Bearer token via mcp-remote. | Sifflet API token (Viewer role for most tools, Editor for monitor generation) and backend URL as environment variables. | | Access | Configurable | Read and write | | Tools listed | 15 named in the evidence | 10 named in the evidence | | Tool definitions read at | none found on 10 Nov 2026 | [raw.githubusercontent.com/siffletdata/sifflet-mcp/1f390ca06d7919214bfe067991539502fdb4b9fc/src/sifflet_mcp/server.py](https://raw.githubusercontent.com/siffletdata/sifflet-mcp/1f390ca06d7919214bfe067991539502fdb4b9fc/src/sifflet_mcp/server.py) | | Context weight | Not measured: no published tool-definition text was read on 10 Nov 2026 | 662 tokens for 10 tool definitions (10 with a description), measured from the source code read | | Status | Exists on 10 Nov 2026 as a hosted first-party server; tool names are partial (those named in the docs), full tool definitions were not published as text, so no schema file was saved. | Exists on 10 Nov 2026; Sifflet docs page (updated 1 Oct 2026) points to the siffletdata/sifflet-mcp GitHub repository, MIT license, PyPI 0.1.7 released 3 Aug 2026; listed after the owner asked for one re-check (the docs host docs.sifflet.io did not load on 10 Nov 2026, docs.siffletdata.com did). | | Read | 10 Nov 2026 | 10 Nov 2026 | The same fourteen fields for both servers, read 10 Nov 2026. ## In one sentence each - [Monte Carlo MCP Server](/servers/monte-carlo-mcp/): Monte Carlo's hosted observability server uses per-user OAuth or scoped keys, a per-request read-only header and tool allow-lists. - [Sifflet MCP Server](/servers/sifflet-mcp/): Sifflet's local server covers asset search, incidents, monitors as code and downstream impact from one Sifflet API token. ## The read facts, topic by topic | Topic | Monte Carlo MCP Server | Sifflet MCP Server | | --- | --- | --- | | Publisher and releases | The server is first-party, run by Monte Carlo at mcp.getmontecarlo.com/mcp, not distributed as a package. No version number or public release date is published for the hosted server; the docs page carries an updatedAt of 2 Nov 2026. | The repository is siffletdata/sifflet-mcp under the MIT license; its pyproject names version 0.1.7. PyPI lists sifflet-mcp 0.1.7, uploaded 3 Aug 2026; the previous releases are 0.1.6 (3 Sep 2025) and 0.1.5 (2 Jun 2025). Sifflet's own documentation page for the MCP server points to the GitHub repository for installation. | | Clients, auth and scoping | Monte Carlo says connected agents can investigate alerts, explore assets and lineage, and create and manage monitors. OAuth lets any HTTP-transport MCP client connect with no API keys, and the connector is listed in the Claude Connectors Directory. An Editor role or above is required to use the server. An agent_observability toolset adds get_agent_metadata, get_agent_conversation and get_agent_trace for inspecting AI agents. | The README documents setup for Cursor and Claude Desktop through uvx sifflet-mcp@latest. | | Access and safety | An x-mcp-readonly header hides write tools such as update_alert and create_or_update_* monitors and blocks calls to them. Per-request x-mcp-tools allow-lists narrow tools and are enforced on call, not only on listing. API-key access uses MCP-server-specific scoped keys; standard API keys do not work. The MCP server can be restricted by IP using Network Access Control Lists. External security audits and penetration tests are done at least annually, covering MCP surfaces. | A Viewer-role token is enough for most tools; get_monitor_code_by_description needs the Editor role on the domain. Two tools change state: close_incident_by_id and open_incident_by_id patch an incident's status. None found: An audit log or per-call permission prompt described for the MCP server in the README or the docs page. | | What it reaches | None found: A list of warehouses and BI systems on the MCP pages (the server fronts the Monte Carlo account's monitored assets). | The search_asset tool filters by asset type TABLE_AND_VIEW, PIPELINE, DASHBOARD or ML_MODEL. | | Lineage and impact | Monte Carlo's MCP server includes a get_asset_lineage tool, used to see how data flows into a table. None found: Documentation that get_asset_lineage returns column-level lineage, transformation SQL, BI fields or metrics; the MCP pages name only asset lineage. | Sifflet MCP exposes get_downstream_assets_of_asset, which returns all downstream assets of an asset by URN. Sifflet says the server can start from an incident and trace the downstream assets affected. | | Definitions, ownership and rules | None found: Glossary terms, definitions or classification tools in the Monte Carlo MCP server docs; the tools are alerts, monitors, assets, lineage and data operations. Governance-like context comes from use cases and critical-table groupings (get_use_cases, get_use_case_tables) in the extended toolset. | Sifflet says asset search returns schema, owners, tags and metadata for tables, views and dashboards. | | Freshness and upkeep | None found: A freshness statement on the MCP pages (the server reads live alert and monitor state from the Monte Carlo account). | The server reads live from the Sifflet API at call time with the configured backend URL; no cached copy is described. | | Profile | Monte Carlo also publishes an Agent Toolkit (mc-agent-toolkit) of skills for monitoring and code-change validation, separate from the server. | none found on 10 Nov 2026 | Each cell joins the facts read for that topic; the full list with sources is on each entry. Read 10 Nov 2026. ## Tool names, as published ### Monte Carlo MCP Server `search`, `get_table`, `get_asset_lineage`, `get_alerts`, `alert_assessment`, `update_alert`, `get_monitors`, `create_or_update_table_monitor`, `get_use_cases`, `get_use_case_tables`, `get_use_case_table_summary`, `get_agent_metadata`, `get_agent_conversation`, `get_agent_trace`, `create_agent_metric_monitor` ### Sifflet MCP Server `asset_by_urn`, `search_asset`, `search_incidents`, `get_incident_by_issue_number`, `get_incident_scope_by_issue_number`, `get_monitor_details_by_id`, `close_incident_by_id`, `open_incident_by_id`, `get_monitor_code_by_description`, `get_downstream_assets_of_asset` ## Read next - [Monte Carlo MCP Server, the full entry](/servers/monte-carlo-mcp/) - [Sifflet MCP Server, the full entry](/servers/sifflet-mcp/) - [Every side-by-side](/compare/) ## Sources - docs.getmontecarlo.com/docs/mcp-server: https://docs.getmontecarlo.com/docs/mcp-server (read 10 Nov 2026) - docs.getmontecarlo.com/docs/mcp-server-reference-and-operations: https://docs.getmontecarlo.com/docs/mcp-server-reference-and-operations (read 10 Nov 2026) - docs.getmontecarlo.com/docs/ai-security: https://docs.getmontecarlo.com/docs/ai-security (read 10 Nov 2026) - github.com/siffletdata/sifflet-mcp/blob/1f390ca06d7919214bfe067991539502fdb4b9fc/src/sifflet_mcp/server.py: https://github.com/siffletdata/sifflet-mcp/blob/1f390ca06d7919214bfe067991539502fdb4b9fc/src/sifflet_mcp/server.py (read 10 Nov 2026) - github.com/siffletdata/sifflet-mcp/blob/main/README.md: https://github.com/siffletdata/sifflet-mcp/blob/main/README.md (read 10 Nov 2026) - raw.githubusercontent.com/siffletdata/sifflet-mcp/main/pyproject.toml: https://raw.githubusercontent.com/siffletdata/sifflet-mcp/main/pyproject.toml (read 10 Nov 2026) - pypi.org/project/sifflet-mcp: https://pypi.org/project/sifflet-mcp/ (read 10 Nov 2026) - docs.siffletdata.com/docs/sifflet-mcp-server.md: https://docs.siffletdata.com/docs/sifflet-mcp-server.md (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/compare/monte-carlo-mcp-vs-sifflet-mcp/ --- # Power BI Modeling MCP Server and Tableau MCP, field by field: 21 and 40 tools named SIDE BY SIDE, read 10 Nov 2026 **The short answer.** Power BI Modeling MCP Server (Microsoft, remote and local) names 21 tools; Tableau MCP (Tableau, remote and local) names 40. Versions read on 10 Nov 2026: 1.0.1 and 6.0.4. The page puts the same fields in the same rows and draws no conclusion. | Field | Power BI Modeling MCP Server | Tableau MCP | | --- | --- | --- | | Data tool | [Power BI](/mcp/power-bi/) | [Tableau](/mcp/tableau/) | | Publisher | Microsoft | Tableau (Salesforce) | | Kind | Official | Official | | Repository or docs | [github.com/microsoft/powerbi-modeling-mcp](https://github.com/microsoft/powerbi-modeling-mcp) | [github.com/tableau/tableau-mcp](https://github.com/tableau/tableau-mcp) | | Version or commit | 1.0.1 | 6.0.4 | | Last release | 6 Nov 2026 | 9 Nov 2026 | | Transport | Remote and local | Remote and local | | Auth | Local: Microsoft Entra interactive sign-in (default) or service principal (client secret or certificate). Hosted: Entra ID as the signed-in user. | Hosted: OAuth 2.1 with each user's own Tableau Cloud identity. Self-hosted: PAT (default), direct-trust (Connected Apps), OAuth, passthrough, or unified access tokens. | | Access | Configurable | Configurable | | Tools listed | 21 named in the evidence | 40 named in the evidence | | Tool definitions read at | [raw.githubusercontent.com/microsoft/powerbi-modeling-mcp/main/README.md](https://raw.githubusercontent.com/microsoft/powerbi-modeling-mcp/main/README.md) | [raw.githubusercontent.com/tableau/tableau-mcp/main/src/tools/web/getDatasourceMetadata/getDatasourceMetadata.ts](https://raw.githubusercontent.com/tableau/tableau-mcp/main/src/tools/web/getDatasourceMetadata/getDatasourceMetadata.ts) | | Context weight | 387 tokens for 21 tool definitions (21 with a description), measured from the README read | Not measured: the source read on 10 Nov 2026 holds the definitions of 2 of the roughly 40 Tableau MCP tools (get-datasource-metadata and query-datasource), and one of those descriptions is imported from a file that was not read, so a measurement would cover a small part of the server. | | Status | Exists on 10 Nov 2026. Microsoft documents two Power BI MCP servers: the Authoring server (local GA over stdio from microsoft/powerbi-modeling-mcp, hosted preview at api.fabric.microsoft.com/v1/mcp/powerbi/authoring and Fabric IQ for answering questions over governed data; an earlier consumption MCP remains documented. Hosted authoring tool names are not published, so the tool list is the local server's. | Exists on 10 Nov 2026 as tableau/tableau-mcp (npm @tableau/mcp-server) and as a hosted service at mcp.tableau.com; the schema file holds two of the tool source files (get-datasource-metadata, query-datasource); the full tool list comes from the docs intro page. | | Read | 10 Nov 2026 | 10 Nov 2026 | The same fourteen fields for both servers, read 10 Nov 2026. ## In one sentence each - [Power BI Modeling MCP Server](/servers/powerbi-modeling-mcp/): Microsoft's authoring server reads and writes semantic-model objects (tables, columns, DAX measures, relationships, RLS roles, TMDL) under the user's Fabric permissions, with a read-only switch. - [Tableau MCP](/servers/tableau-mcp/): Tableau's server exposes data source metadata with calculation strings and Pulse metric definitions per signed-in user, plus, on Tableau+, a knowledge tool for lineage and impact. ## The read facts, topic by topic | Topic | Power BI Modeling MCP Server | Tableau MCP | | --- | --- | --- | | Publisher and releases | Microsoft publishes the Power BI Authoring MCP Server in its own microsoft GitHub organization and documents it on Microsoft Learn. The latest GitHub release is v1.0.1 (previous v1.0.0, then beta tags), timestamped 6 Nov 2026. The npm package @microsoft/powerbi-modeling-mcp has latest 1.0.1 published 6 Nov 2026, and 1.0.0 on 25 Sep 2026. The local server is generally available and the hosted endpoint is in preview. | Tableau publishes tableau-mcp as its official MCP server, and the repo README carries the 'Tableau Supported' support-level badge. The latest GitHub release is v6.0.4 (listed first on the releases page). The npm package @tableau/mcp-server has dist-tag latest 6.0.4, published 9 Nov 2026, under the Apache-2.0 license. The README badge states the support level as Tableau Supported. | | Clients, auth and scoping | The hosted server prompts for Microsoft Entra sign-in on first tool call and then acts with the user's permissions. Fabric IQ, Microsoft's consumption server, answers questions over semantic models and respects permissions, row-level security and object-level security. Microsoft states that clients operate under the user's Fabric RBAC permissions. | The hosted Tableau MCP at mcp.tableau.com uses OAuth 2.1 so each user signs in as themselves and per-user Tableau permissions are enforced. Authentication to the Tableau REST APIs is selectable between pat (default), direct-trust, oauth and passthrough. The Tableau docs state the authenticated user must have access to the published data sources the agent will use. | | Access and safety | The --readonly option is described as safe mode that prevents any write operations to the semantic model; --readwrite is the default with a confirmation prompt. Blocking access requires disabling the XMLA endpoint because no tenant admin setting specifically blocks this server. With only Build permission the server can only run DAX queries; Write permission is needed to change model objects. | Tools can be limited with INCLUDE_TOOLS or EXCLUDE_TOOLS environment variables (comma-separated tool or group names). Mutating flow tools (run-flow, run-flow-task, cancel-flow-run) are described as preview-and-confirm tools, and delete-content is admin-only and two-phase. The query-datasource tool is annotated readOnlyHint true and destructiveHint false in source. Telemetry defaults to the noop provider. | | What it reaches | The local server connects to semantic models in Power BI Desktop, Fabric workspaces and Power BI Project (PBIP) files; --compatibility Full also targets Analysis Services databases. Hosted uses Streamable HTTP and Entra ID; local uses stdio, with Desktop, PBIP files, transactions and traces only on local. | The server works with Tableau Cloud and Tableau Server of any edition, with some tools gated by SKU. Tools span published data sources, workbooks, views, custom views, projects, Prep flows, Pulse metrics, extract refresh tasks, users and Admin Insights. The server speaks stdio by default and HTTP as an alternative transport. | | Lineage and impact | The server exposes model-object tools for tables, columns, measures, relationships, partitions, hierarchies, calculation groups and named expressions (model definition level). Database_operations can import and export TMDL folders and deploy to Fabric, giving agents the model definition as code. None found: A lineage tool in the Power BI Authoring MCP tool table (no upstream-table, column-lineage or report-dependency tool). | Tableau's get-datasource-metadata tool returns table relationships, field descriptions, roles and types, calculation strings and parameters for a data source (table, column and calculation-definition level, not transformation lineage). The describe-flow tool summarizes a Tableau Prep flow's design including inputs, outputs, steps, lineage and connections, on all SKUs. On Tableau+ only, query-knowledge-context queries governed definitions, relationships, lineage, impact and graph sources through one read-only tool. The server's get-datasource-metadata tool enriches VizQL Data Service results with Tableau's Metadata API and only returns the richer descriptions for published data sources. | | Definitions, ownership and rules | Tools manage security roles with row-level security, perspectives, cultures and object translations, which are governance and meaning objects inside the model. The server can evaluate a model against a set of modeling practice rules and generate documentation for an existing model. | Tableau MCP serves Pulse metric definitions and the Pulse insights engine to agents so metrics are answered from governed definitions. On Tableau+ only, manage-knowledge-context creates, updates or deletes customer-governed semantic context, and inspect-knowledge-context reports graph status and improvement suggestions. The intro page positions the server on reusing curated data models and metadata so agent answers are grounded in pre-built data semantics. | | Freshness and upkeep | The server connects to semantic models through the XMLA endpoint, so it reads and writes the live model rather than a collected copy. None found: Scheduled metadata collection or lineage refresh in the Power BI Authoring MCP README or Learn page. | Query-datasource runs VizQL Data Service queries against published data sources at call time, so answers reflect live data rather than a collected copy. None found: Any scheduled or event-driven metadata collection in the Tableau MCP docs intro page; the server calls Tableau REST, Metadata and VizQL APIs per request. | | Profile | Use of the software is subject to the Power BI Authoring MCP Server EULA, and NPX installs require explicit acceptance. Microsoft advises the hosted server for Fabric workspaces and not registering both local and remote at once. | The docs site footer credits Salesforce, Inc. as the copyright holder of Tableau MCP. The package can run via npx (Node 22.7.5 or later), Docker, Heroku, or as Node single-executable binaries; a separate Tableau Desktop authoring server is not in the npm package. | Each cell joins the facts read for that topic; the full list with sources is on each entry. Read 10 Nov 2026. ## Tool names, as published ### Power BI Modeling MCP Server `connection_operations`, `database_operations`, `transaction_operations`, `model_operations`, `table_operations`, `column_operations`, `measure_operations`, `relationship_operations`, `dax_query_operations`, `trace_operations`, `partition_operations`, `user_hierarchy_operations`, `calculation_group_operations`, `security_role_operations`, `perspective_operations`, `named_expression_operations`, `function_operations`, `culture_operations`, `object_translation_operations`, `calendar_operations`, `query_group_operations` ### Tableau MCP `list-datasources`, `list-workbooks`, `list-projects`, `list-views`, `list-custom-views`, `get-datasource-metadata`, `get-workbook`, `download-workbook`, `get-flow`, `describe-flow`, `get-flow-task`, `list-flows`, `list-flow-runs`, `list-flow-tasks`, `run-flow`, `run-flow-task`, `cancel-flow-run`, `delete-content`, `get-view-data`, `get-view-image`, `get-custom-view-data`, `get-custom-view-image`, `query-datasource`, `list-all-pulse-metric-definitions`, `list-pulse-metric-definitions-from-definition-ids`, `list-pulse-metrics-from-metric-definition-id`, `list-pulse-metrics-from-metric-ids`, `list-pulse-metric-subscriptions`, `generate-pulse-metric-value-insight-bundle`, `generate-pulse-insight-brief`, `search-content`, `list-extract-refresh-tasks`, `update-cloud-extract-refresh-task`, `list-users`, `update-user`, `query-admin-insights`, `query-knowledge-context`, `inspect-knowledge-context`, `manage-knowledge-context`, `scaffold-data-app` ## Read next - [Power BI Modeling MCP Server, the full entry](/servers/powerbi-modeling-mcp/) - [Tableau MCP, the full entry](/servers/tableau-mcp/) - [Every side-by-side](/compare/) ## Sources - learn.microsoft.com/en-us/power-bi/developer/mcp/power-bi-authoring-mcp: https://learn.microsoft.com/en-us/power-bi/developer/mcp/power-bi-authoring-mcp (read 10 Nov 2026) - github.com/microsoft/powerbi-modeling-mcp/releases: https://github.com/microsoft/powerbi-modeling-mcp/releases (read 10 Nov 2026) - registry.npmjs.org/@microsoft/powerbi-modeling-mcp: https://registry.npmjs.org/@microsoft/powerbi-modeling-mcp (read 10 Nov 2026) - raw.githubusercontent.com/microsoft/powerbi-modeling-mcp/main/README.md: https://raw.githubusercontent.com/microsoft/powerbi-modeling-mcp/main/README.md (read 10 Nov 2026) - learn.microsoft.com/en-us/power-bi/developer/mcp/mcp-servers-overview: https://learn.microsoft.com/en-us/power-bi/developer/mcp/mcp-servers-overview (read 10 Nov 2026) - tableau.github.io/tableau-mcp/docs/intro: https://tableau.github.io/tableau-mcp/docs/intro (read 10 Nov 2026) - github.com/tableau/tableau-mcp/releases: https://github.com/tableau/tableau-mcp/releases (read 10 Nov 2026) - registry.npmjs.org/@tableau/mcp-server: https://registry.npmjs.org/@tableau/mcp-server (read 10 Nov 2026) - raw.githubusercontent.com/tableau/tableau-mcp/main/README.md: https://raw.githubusercontent.com/tableau/tableau-mcp/main/README.md (read 10 Nov 2026) - raw.githubusercontent.com/tableau/tableau-mcp/main/src/tools/web/getDatasourceMetadata/getDatasourceMetadata.ts: https://raw.githubusercontent.com/tableau/tableau-mcp/main/src/tools/web/getDatasourceMetadata/getDatasourceMetadata.ts (read 10 Nov 2026) - github.com/tableau/tableau-mcp: https://github.com/tableau/tableau-mcp (read 10 Nov 2026) - tableau.github.io/tableau-mcp/docs/configuration/mcp-config/env-vars: https://tableau.github.io/tableau-mcp/docs/configuration/mcp-config/env-vars (read 10 Nov 2026) - tableau.github.io/tableau-mcp/docs/configuration/mcp-config/authentication: https://tableau.github.io/tableau-mcp/docs/configuration/mcp-config/authentication (read 10 Nov 2026) - raw.githubusercontent.com/tableau/tableau-mcp/main/src/tools/web/queryDatasource/queryDatasource.ts: https://raw.githubusercontent.com/tableau/tableau-mcp/main/src/tools/web/queryDatasource/queryDatasource.ts (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/compare/powerbi-modeling-mcp-vs-tableau-mcp/ --- # Looker tools in MCP Toolbox for Databases and Tableau MCP, field by field: 55 and 40 tools named SIDE BY SIDE, read 10 Nov 2026 **The short answer.** Looker tools in MCP Toolbox for Databases (Google, remote and local) names 55 tools; Tableau MCP (Tableau, remote and local) names 40. Versions read on 10 Nov 2026: v1.16.0 and 6.0.4. The page puts the same fields in the same rows and draws no conclusion. | Field | Looker tools in MCP Toolbox for Databases | Tableau MCP | | --- | --- | --- | | Data tool | [Looker](/mcp/looker/) | [Tableau](/mcp/tableau/) | | Publisher | Google (googleapis/mcp-toolbox) | Tableau (Salesforce) | | Kind | Official | Official | | Repository or docs | [github.com/googleapis/mcp-toolbox](https://github.com/googleapis/mcp-toolbox) | [github.com/tableau/tableau-mcp](https://github.com/tableau/tableau-mcp) | | Version or commit | V1.16.0 | 6.0.4 | | Last release | 5 Nov 2026 | 9 Nov 2026 | | Transport | Remote and local | Remote and local | | Auth | Looker API client_id and client_secret in the Looker source config, or use_client_oauth to pass the caller's OAuth token (default false). | Hosted: OAuth 2.1 with each user's own Tableau Cloud identity. Self-hosted: PAT (default), direct-trust (Connected Apps), OAuth, passthrough, or unified access tokens. | | Access | Configurable | Configurable | | Tools listed | 55 named in the evidence | 40 named in the evidence | | Tool definitions read at | [mcp-toolbox.dev/llms.txt](https://mcp-toolbox.dev/llms.txt) | [raw.githubusercontent.com/tableau/tableau-mcp/main/src/tools/web/getDatasourceMetadata/getDatasourceMetadata.ts](https://raw.githubusercontent.com/tableau/tableau-mcp/main/src/tools/web/getDatasourceMetadata/getDatasourceMetadata.ts) | | Context weight | 1,475 tokens for 55 tool definitions (55 with a description), measured from the docs index read | Not measured: the source read on 10 Nov 2026 holds the definitions of 2 of the roughly 40 Tableau MCP tools (get-datasource-metadata and query-datasource), and one of those descriptions is imported from a file that was not read, so a measurement would cover a small part of the server. | | Status | Exists on 10 Nov 2026 as the looker source type in MCP Toolbox for Databases; the repo URL github.com/googleapis/genai-toolbox now titles itself googleapis/mcp-toolbox. 55 looker-* tools are listed in the Toolbox llms.txt index; schema file holds the one-line descriptions only. | Exists on 10 Nov 2026 as tableau/tableau-mcp (npm @tableau/mcp-server) and as a hosted service at mcp.tableau.com; the schema file holds two of the tool source files (get-datasource-metadata, query-datasource); the full tool list comes from the docs intro page. | | Read | 10 Nov 2026 | 10 Nov 2026 | The same fourteen fields for both servers, read 10 Nov 2026. ## In one sentence each - [Looker tools in MCP Toolbox for Databases](/servers/looker-mcp-toolbox/): Google's Toolbox exposes Looker models, explores, dimensions, measures and LookML project files, plus instance health and usage tools. - [Tableau MCP](/servers/tableau-mcp/): Tableau's server exposes data source metadata with calculation strings and Pulse metric definitions per signed-in user, plus, on Tableau+, a knowledge tool for lineage and impact. ## The read facts, topic by topic | Topic | Looker tools in MCP Toolbox for Databases | Tableau MCP | | --- | --- | --- | | Publisher and releases | Looker's tools ship in MCP Toolbox for Databases, published under Google's googleapis GitHub organization. The latest GitHub release is v1.16.0, timestamped 5 Nov 2026; the releases page title shows the repo as googleapis/mcp-toolbox. The release list shows v1.16.0, v1.15.0, v1.14.0 and v1.13.1 as the four most recent tags. | Tableau publishes tableau-mcp as its official MCP server, and the repo README carries the 'Tableau Supported' support-level badge. The latest GitHub release is v6.0.4 (listed first on the releases page). The npm package @tableau/mcp-server has dist-tag latest 6.0.4, published 9 Nov 2026, under the Apache-2.0 license. The README badge states the support level as Tableau Supported. | | Clients, auth and scoping | The Looker source can pass the caller's OAuth token instead of one service credential via use_client_oauth, which defaults to false. The default source config authenticates with a single client_id and client_secret, so identity is that API user unless OAuth is enabled. The source can hide models, explores and fields from agents with show_hidden_models, show_hidden_explores and show_hidden_fields (all default true). | The hosted Tableau MCP at mcp.tableau.com uses OAuth 2.1 so each user signs in as themselves and per-user Tableau permissions are enforced. Authentication to the Tableau REST APIs is selectable between pat (default), direct-trust, oauth and passthrough. The Tableau docs state the authenticated user must have access to the published data sources the agent will use. | | Access and safety | Toolbox documents a Read-Only Tools page covering engine-level protocol locks, tool suppression and MCP annotations. The Looker toolset includes write tools such as creating or deleting LookML project files and switching Looker into dev mode, so access is not read-only by default. Toolbox documents screening traffic with Google Cloud Model Armor to protect against prompt injection and sensitive data leakage. Toolbox includes OpenTelemetry observability built in. | Tools can be limited with INCLUDE_TOOLS or EXCLUDE_TOOLS environment variables (comma-separated tool or group names). Mutating flow tools (run-flow, run-flow-task, cancel-flow-run) are described as preview-and-confirm tools, and delete-content is admin-only and two-phase. The query-datasource tool is annotated readOnlyHint true and destructiveHint false in source. Telemetry defaults to the noop provider. | | What it reaches | Looker can be deployed in the cloud, on GCP, or on premises, and the same Toolbox covers Google Cloud databases such as AlloyDB, BigQuery, Cloud SQL, Spanner and Firestore. Looker connection tools list databases, schemas, tables and columns of the warehouse connections configured in Looker. Toolbox runs as a local stdio MCP server through npx @toolbox-sdk/server with the --stdio flag. | The server works with Tableau Cloud and Tableau Server of any edition, with some tools gated by SKU. Tools span published data sources, workbooks, views, custom views, projects, Prep flows, Pulse metrics, extract refresh tasks, users and Admin Insights. The server speaks stdio by default and HTTP as an alternative transport. | | Lineage and impact | Looker-get-explore returns explore metadata including always_filter and conditionally_filter constraints, tags and hidden status. Separate tools return dimensions, measures, filters and parameters of an explore, and the columns of connection tables. Looker-get-project-file returns the contents of a LookML file, so an agent can read the transformation and field logic written in LookML. None found: A lineage tool among the 55 looker-* tools in the Toolbox llms.txt index (no column-lineage or upstream-table tool is named). | Tableau's get-datasource-metadata tool returns table relationships, field descriptions, roles and types, calculation strings and parameters for a data source (table, column and calculation-definition level, not transformation lineage). The describe-flow tool summarizes a Tableau Prep flow's design including inputs, outputs, steps, lineage and connections, on all SKUs. On Tableau+ only, query-knowledge-context queries governed definitions, relationships, lineage, impact and graph sources through one read-only tool. The server's get-datasource-metadata tool enriches VizQL Data Service results with Tableau's Metadata API and only returns the richer descriptions for published data sources. | | Definitions, ownership and rules | The Looker source page describes Looker as providing a semantic layer to facilitate querying. Looker-health-analyze reports unused explores, joins and fields, defining unused as not used as a field or filter in the past 90 days in production. Looker-conversational-analytics uses the Conversational Analytics API to analyze Looker data, and agents can be created and managed with looker-create-agent. | Tableau MCP serves Pulse metric definitions and the Pulse insights engine to agents so metrics are answered from governed definitions. On Tableau+ only, manage-knowledge-context creates, updates or deletes customer-governed semantic context, and inspect-knowledge-context reports graph status and improvement suggestions. The intro page positions the server on reusing curated data models and metadata so agent answers are grounded in pre-built data semantics. | | Freshness and upkeep | Looker-health-pulse performs health checks on a Looker instance, including database connections and dashboard performance, at call time. None found: Scheduled or event-driven metadata collection for Looker in the Toolbox docs; tools call the Looker API per request. | Query-datasource runs VizQL Data Service queries against published data sources at call time, so answers reflect live data rather than a collected copy. None found: Any scheduled or event-driven metadata collection in the Tableau MCP docs intro page; the server calls Tableau REST, Metadata and VizQL APIs per request. | | Profile | The repo was renamed: the genai-toolbox URL now serves googleapis/mcp-toolbox, and prebuilt tools exist for Looker alongside many Google databases. | The docs site footer credits Salesforce, Inc. as the copyright holder of Tableau MCP. The package can run via npx (Node 22.7.5 or later), Docker, Heroku, or as Node single-executable binaries; a separate Tableau Desktop authoring server is not in the npm package. | Each cell joins the facts read for that topic; the full list with sources is on each entry. Read 10 Nov 2026. ## Tool names, as published ### Looker tools in MCP Toolbox for Databases `looker-add-dashboard-element`, `looker-add-dashboard-filter`, `looker-conversational-analytics`, `looker-create-agent`, `looker-create-dashboard-layout`, `looker-create-git-branch`, `looker-create-project-directory`, `looker-create-project-file`, `looker-delete-agent`, `looker-delete-git-branch`, `looker-delete-project-directory`, `looker-delete-project-file`, `looker-dev-mode`, `looker-generate-embed-url`, `looker-get-agent`, `looker-get-connection-databases`, `looker-get-connection-schemas`, `looker-get-connection-table-columns`, `looker-get-connection-tables`, `looker-get-connections`, `looker-get-dashboard`, `looker-get-dashboards`, `looker-get-dimensions`, `looker-get-explore`, `looker-get-explores`, `looker-get-field-value-suggestions`, `looker-get-filters`, `looker-get-git-branch`, `looker-get-looks`, `looker-get-measures`, `looker-get-models`, `looker-get-parameters`, `looker-get-project-directories`, `looker-get-project-file`, `looker-get-project-files`, `looker-get-projects`, `looker-health-analyze`, `looker-health-pulse`, `looker-health-vacuum`, `looker-import-dashboard-from-lookml`, `looker-list-agents`, `looker-list-git-branches`, `looker-make-dashboard`, `looker-make-look`, `looker-query`, `looker-query-sql`, `looker-query-url`, `looker-run-dashboard`, `looker-run-look`, `looker-switch-git-branch`, `looker-update-agent`, `looker-update-dashboard-element`, `looker-update-dashboard-layout-component`, `looker-update-project-file`, `looker-validate-project` ### Tableau MCP `list-datasources`, `list-workbooks`, `list-projects`, `list-views`, `list-custom-views`, `get-datasource-metadata`, `get-workbook`, `download-workbook`, `get-flow`, `describe-flow`, `get-flow-task`, `list-flows`, `list-flow-runs`, `list-flow-tasks`, `run-flow`, `run-flow-task`, `cancel-flow-run`, `delete-content`, `get-view-data`, `get-view-image`, `get-custom-view-data`, `get-custom-view-image`, `query-datasource`, `list-all-pulse-metric-definitions`, `list-pulse-metric-definitions-from-definition-ids`, `list-pulse-metrics-from-metric-definition-id`, `list-pulse-metrics-from-metric-ids`, `list-pulse-metric-subscriptions`, `generate-pulse-metric-value-insight-bundle`, `generate-pulse-insight-brief`, `search-content`, `list-extract-refresh-tasks`, `update-cloud-extract-refresh-task`, `list-users`, `update-user`, `query-admin-insights`, `query-knowledge-context`, `inspect-knowledge-context`, `manage-knowledge-context`, `scaffold-data-app` ## Read next - [Looker tools in MCP Toolbox for Databases, the full entry](/servers/looker-mcp-toolbox/) - [Tableau MCP, the full entry](/servers/tableau-mcp/) - [Every side-by-side](/compare/) ## Sources - raw.githubusercontent.com/googleapis/genai-toolbox/main/README.md: https://raw.githubusercontent.com/googleapis/genai-toolbox/main/README.md (read 10 Nov 2026) - github.com/googleapis/mcp-toolbox/releases: https://github.com/googleapis/mcp-toolbox/releases (read 10 Nov 2026) - mcp-toolbox.dev/integrations/looker/tools/looker-get-explore: https://mcp-toolbox.dev/integrations/looker/tools/looker-get-explore/ (read 10 Nov 2026) - mcp-toolbox.dev/llms.txt: https://mcp-toolbox.dev/llms.txt (read 10 Nov 2026) - mcp-toolbox.dev/integrations/looker/source: https://mcp-toolbox.dev/integrations/looker/source/ (read 10 Nov 2026) - mcp-toolbox.dev/integrations/looker/tools/looker-health-analyze: https://mcp-toolbox.dev/integrations/looker/tools/looker-health-analyze/ (read 10 Nov 2026) - github.com/googleapis/mcp-toolbox: https://github.com/googleapis/mcp-toolbox (read 10 Nov 2026) - tableau.github.io/tableau-mcp/docs/intro: https://tableau.github.io/tableau-mcp/docs/intro (read 10 Nov 2026) - github.com/tableau/tableau-mcp/releases: https://github.com/tableau/tableau-mcp/releases (read 10 Nov 2026) - registry.npmjs.org/@tableau/mcp-server: https://registry.npmjs.org/@tableau/mcp-server (read 10 Nov 2026) - raw.githubusercontent.com/tableau/tableau-mcp/main/README.md: https://raw.githubusercontent.com/tableau/tableau-mcp/main/README.md (read 10 Nov 2026) - raw.githubusercontent.com/tableau/tableau-mcp/main/src/tools/web/getDatasourceMetadata/getDatasourceMetadata.ts: https://raw.githubusercontent.com/tableau/tableau-mcp/main/src/tools/web/getDatasourceMetadata/getDatasourceMetadata.ts (read 10 Nov 2026) - github.com/tableau/tableau-mcp: https://github.com/tableau/tableau-mcp (read 10 Nov 2026) - tableau.github.io/tableau-mcp/docs/configuration/mcp-config/env-vars: https://tableau.github.io/tableau-mcp/docs/configuration/mcp-config/env-vars (read 10 Nov 2026) - tableau.github.io/tableau-mcp/docs/configuration/mcp-config/authentication: https://tableau.github.io/tableau-mcp/docs/configuration/mcp-config/authentication (read 10 Nov 2026) - raw.githubusercontent.com/tableau/tableau-mcp/main/src/tools/web/queryDatasource/queryDatasource.ts: https://raw.githubusercontent.com/tableau/tableau-mcp/main/src/tools/web/queryDatasource/queryDatasource.ts (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/compare/looker-mcp-toolbox-vs-tableau-mcp/ --- # Cube MCP server and dbt MCP Server, field by field: 30 and 52 tools named SIDE BY SIDE, read 10 Nov 2026 **The short answer.** Cube MCP server (Cube, remote (hosted)) names 30 tools; dbt MCP Server (dbt Labs, remote and local) names 52. Versions read on 10 Nov 2026: no version number published and 2.10.1. The page puts the same fields in the same rows and draws no conclusion. | Field | Cube MCP server | dbt MCP Server | | --- | --- | --- | | Data tool | [Cube](/mcp/cube/) | [dbt](/mcp/dbt/) | | Publisher | Cube | dbt Labs | | Kind | Official | Official | | Repository or docs | [docs.cube.dev/docs/integrations/mcp-server](https://docs.cube.dev/docs/integrations/mcp-server) | [github.com/dbt-labs/dbt-mcp](https://github.com/dbt-labs/dbt-mcp) | | Version or commit | Hosted service (no versioned release) | 2.10.1 | | Last release | none found on 10 Nov 2026 | 3 Nov 2026 | | Transport | Remote (hosted) | Remote and local | | Auth | OAuth Authorization Code + PKCE (scope mcp-agent-access, public client cube-mcp-client), or a Cube API key exchanged for a token (client_id cube-api-key). | dbt service/personal token and environment IDs for platform tools; OAuth or token per docs; local dbt project for CLI tools. | | Access | Read and write | Configurable | | Tools listed | 30 named in the evidence | 52 named in the evidence | | Tool definitions read at | [docs.cube.dev/docs/integrations/mcp-server.md](https://docs.cube.dev/docs/integrations/mcp-server.md) | [raw.githubusercontent.com/dbt-labs/dbt-mcp/main/README.md](https://raw.githubusercontent.com/dbt-labs/dbt-mcp/main/README.md) | | Context weight | 1,093 tokens for 30 tool definitions (30 with a description), measured from the docs page read | 1,061 tokens for 63 tool definitions (63 with a description), measured from the README read | | Status | Exists on 10 Nov 2026 as a Cube-hosted endpoint cubecloud.dev/mcp (or your own console domain). The page says it exposes 30 tools; the tables list 28 names, all captured here. | Exists, v2.10.1 on 3 Nov 2026 (PyPI and GitHub). | | Read | 10 Nov 2026 | 10 Nov 2026 | The same fourteen fields for both servers, read 10 Nov 2026. ## In one sentence each - [Cube MCP server](/servers/cube-mcp/): Cube's hosted server searches the semantic layer, runs SQL with row-level security as the signed-in user and edits the semantic model on a branch. - [dbt MCP Server](/servers/dbt-mcp/): dbt Labs' server combines the dbt CLI, the Discovery API with model and column lineage, the Semantic Layer and SQL tools on dbt Platform. ## The read facts, topic by topic | Topic | Cube MCP server | dbt MCP Server | | --- | --- | --- | | Publisher and releases | Cube hosts the MCP server and documents it on docs.cube.dev, available on all plans. No versioned release is published; one endpoint cubecloud.dev/mcp serves every account, deployment and region. | dbt Labs publishes dbt-mcp; the latest GitHub release is v2.10.1, whose release timestamp on the page is 3 Nov 2026 (PyPI also 3 Nov 2026). dbt Labs documents the MCP server on docs.getdbt.com and ships it as self-hosted or remote. | | Clients, auth and scoping | Every tool runs as the authenticated user, and queries respect Cube permissions including row-level security. With an API key, a token can act as the key's owner, as another user, or be limited to specific tools. Which tools a user sees depends on their role; Viewer role or higher is needed, and data model editing tools show only for roles allowed to edit the model (Admin and Developer by default). | The dbt MCP server provides AI agents with context of a project in dbt Core, dbt Fusion and dbt Platform. | | Access and safety | Nine tools are annotated destructive and clients that honor annotations always ask before running them. Chat is annotated read-only but runs the Analytics Chat agent, which can create reports or delete folders without confirmation, so the annotation understates what it can do. File writes never touch production: writeDataModelFile and deleteDataModelFile reject any branch that isn't a dev branch. API-key tokens last up to one hour and there is no refresh token. | dbt CLI tools can modify data models and warehouse objects, so the README warns to proceed only if you trust the client. dbt docs say the server stores no production data and is a read-only access layer reading from dbt platform in real time. Configuration uses a dbt service token (DBT_TOKEN) and tool groups can be enabled or disabled by environment variables. | | What it reaches | Documented clients are Claude, ChatGPT, Microsoft 365 Copilot, Cursor and Codex, plus other MCP clients over HTTPS. Admins choose the default deployment and can limit which deployments MCP clients can reach, always intersected with role-based permissions. | Discovery tools cover models, sources, exposures (downstream dashboards, apps) and freshness. | | Lineage and impact | SearchDataModel searches the semantic model by similarity, covering views and their measures and dimensions, and returns name, title, description and type. ListDataModelFiles and readDataModelFile read the raw YAML, JavaScript or Python source of the semantic model, so the definition logic itself is readable. None found: A lineage tool in the Cube MCP tool tables (getBranchDiff and getDataModelChanges compare model versions, not data lineage). | Get_lineage returns a bounded lineage graph with type, depth and direction filters. Column-level lineage tools exist: fusion.get_column_lineage via dbt Platform and get_column_lineage locally. Get_lineage_dev reads lineage from the local manifest.json. | | Definitions, ownership and rules | Model editing tools (startDataModelEdit, writeDataModelFile, commitDataModelChanges) let an agent change definitions on a personal dev branch, merged to the default branch only with a confirm-prompting tool. Query results carry the semantic layer: runQuery runs Cube SQL against views whose names and members come from searchDataModel. | Semantic Layer tools list metrics, dimensions, entities and saved queries. Get_all_models returns model names and descriptions; generate_model_yaml can inherit upstream descriptions. | | Freshness and upkeep | The server can target a dev-mode branch, so agents can query a branch's model before it is merged. GetPreAggregationStatus and buildPreAggregation let an agent check and trigger pre-aggregation builds. | Get_model_health returns run status, test results and upstream source freshness from dbt Platform. | | Profile | Admins need admin privileges or a custom role with the Manage MCP server and connectors permission to enable MCP. | The README publishes an experimental MCP bundle (dbt-mcp.mcpb) with each release. | Each cell joins the facts read for that topic; the full list with sources is on each entry. Read 10 Nov 2026. ## Tool names, as published ### Cube MCP server `listDeployments`, `chat`, `loadQueryResults`, `searchDataModel`, `runQuery`, `readWorkbook`, `createWorkbook`, `createReport`, `readReport`, `updateReport`, `deleteReport`, `manageFolders`, `updateDashboard`, `publishDashboard`, `listDataModelFiles`, `readDataModelFile`, `startDataModelEdit`, `writeDataModelFile`, `deleteDataModelFile`, `getDataModelChanges`, `getBranchDiff`, `getDeploymentEnv`, `getBranchState`, `commitDataModelChanges`, `commitToNewBranch`, `mergeToDefaultBranch`, `discardDataModelChanges`, `switchUserBranch`, `getPreAggregationStatus`, `buildPreAggregation` ### dbt MCP Server `execute_sql`, `text_to_sql`, `get_dimension_values`, `get_dimensions`, `get_entities`, `get_metrics_compiled_sql`, `list_metrics`, `list_saved_queries`, `query_metrics`, `get_all_macros`, `get_all_models`, `get_all_sources`, `get_exposures`, `get_lineage`, `get_mart_models`, `get_model_health`, `get_model_performance`, `get_node_details`, `get_related_models`, `search_public_mesh_models`, `build`, `clone`, `compile`, `docs`, `get_lineage_dev`, `get_node_details_dev`, `list`, `parse`, `run`, `show`, `test`, `cancel_job_run`, `get_job_details`, `get_job_run_artifacts`, `get_job_run_details`, `get_job_run_error`, `list_job_run_artifacts`, `list_jobs`, `list_jobs_runs`, `list_projects`, `retry_job_run`, `trigger_job_run`, `generate_model_yaml`, `generate_source`, `generate_staging_model`, `fusion.compile_sql`, `fusion.get_column_lineage`, `get_column_lineage`, `get_product_doc_pages`, `search_product_docs`, `get_mcp_server_branch`, `get_mcp_server_version` ## Read next - [Cube MCP server, the full entry](/servers/cube-mcp/) - [dbt MCP Server, the full entry](/servers/dbt-mcp/) - [Every side-by-side](/compare/) ## Sources - docs.cube.dev/docs/integrations/mcp-server.md: https://docs.cube.dev/docs/integrations/mcp-server.md (read 10 Nov 2026) - github.com/dbt-labs/dbt-mcp/releases: https://github.com/dbt-labs/dbt-mcp/releases (read 10 Nov 2026) - docs.getdbt.com/docs/dbt-ai/about-mcp: https://docs.getdbt.com/docs/dbt-ai/about-mcp (read 10 Nov 2026) - raw.githubusercontent.com/dbt-labs/dbt-mcp/main/README.md: https://raw.githubusercontent.com/dbt-labs/dbt-mcp/main/README.md (read 10 Nov 2026) - raw.githubusercontent.com/dbt-labs/dbt-mcp/main/.env.example: https://raw.githubusercontent.com/dbt-labs/dbt-mcp/main/.env.example (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/compare/cube-mcp-vs-dbt-mcp/ --- # Airbyte MCP (hosted) and Fivetran MCP Server, field by field: 50 and 53 tools named SIDE BY SIDE, read 10 Nov 2026 **The short answer.** Airbyte MCP (hosted) (Airbyte, remote (hosted)) names 50 tools; Fivetran MCP Server (Fivetran, remote and local) names 53. Versions read on 10 Nov 2026: 0.71.3 and v0.5.1. The page puts the same fields in the same rows and draws no conclusion. | Field | Airbyte MCP (hosted) | Fivetran MCP Server | | --- | --- | --- | | Data tool | [Airbyte](/mcp/airbyte/) | [Fivetran](/mcp/fivetran/) | | Publisher | Airbyte | Fivetran | | Kind | Official | Official | | Repository or docs | [docs.airbyte.com/platform/airbyte-mcp](https://docs.airbyte.com/platform/airbyte-mcp) | [github.com/fivetran/fivetran-mcp](https://github.com/fivetran/fivetran-mcp) | | Version or commit | Hosted service (no versioned release); PyAirbyte package 0.71.3 ships an experimental local airbyte.mcp server | V0.5.1 | | Last release | none found on 10 Nov 2026 | 6 Oct 2026 | | Transport | Remote (hosted) | Remote and local | | Auth | OAuth (Streamable HTTP) as the signed-in Airbyte user, or an Airbyte application for non-browser clients; tools run with that user's Airbyte role. | Fivetran API key and API secret (FIVETRAN_API_KEY, FIVETRAN_API_SECRET) from the Fivetran dashboard; one shared credential, not per-user OAuth. Streamable HTTP mode forwards headers; OAuth resource-server support is scaffolding only. | | Access | Configurable | Configurable | | Tools listed | 50 named in the evidence | 53 named in the evidence | | Tool definitions read at | [docs.airbyte.com/platform/airbyte-mcp/tools.md](https://docs.airbyte.com/platform/airbyte-mcp/tools.md) | [raw.githubusercontent.com/fivetran/fivetran-mcp/main/open-api-definitions/AVAILABLE_ACTIONS.md](https://raw.githubusercontent.com/fivetran/fivetran-mcp/main/open-api-definitions/AVAILABLE_ACTIONS.md) | | Context weight | 870 tokens for 50 tool definitions (50 with a description), measured from the docs page read | 299 tokens for 53 tool definitions (2 with a description), measured from the README read | | Status | Exists on 10 Nov 2026 at mcp.airbyte.com/mcp for Airbyte Cloud and Flex; self-managed Core can run it locally with known issues. A separate PyAirbyte (airbyte.mcp) server and an airbyte-agent-connectors SDK also exist; this file covers the documented Airbyte MCP. | Exists on 10 Nov 2026 as fivetran/fivetran-mcp, installed from git with uvx (the PyPI names fivetran-mcp and mcp-fivetran belong to unrelated projects). Tool set is generated from the OpenAPI manifest, so it mirrors the REST API. | | Read | 10 Nov 2026 | 10 Nov 2026 | The same fourteen fields for both servers, read 10 Nov 2026. ## In one sentence each - [Airbyte MCP (hosted)](/servers/airbyte-mcp/): Airbyte's hosted server lets an agent build, run and troubleshoot pipelines as the signed-in user, with a documented read-only mode; it serves pipeline and catalog state. - [Fivetran MCP Server](/servers/fivetran-mcp/): Fivetran's server mirrors its REST API behind a read-by-default scope with a deny-list, so an agent can inspect connections, schemas and sync status. ## The read facts, topic by topic | Topic | Airbyte MCP (hosted) | Fivetran MCP Server | | --- | --- | --- | | Publisher and releases | Airbyte documents the Airbyte MCP as its own Model Context Protocol server connecting agents to an Airbyte organization with the user's own permissions. The hosted server has no versioned release; its URL is mcp.airbyte.com/mcp. The airbyte package on PyPI (PyAirbyte) is at version 0.71.3, uploaded 26 Oct 2026, and includes an experimental airbyte.mcp server module. The PyAirbyte reference describes airbyte.mcp.server as an experimental MCP server for connector management. | Fivetran publishes the server in its own GitHub organization and describes it as an MCP server for interacting with a Fivetran environment. The latest GitHub release is v0.5.1, timestamped 6 Oct 2026 on the releases page. The README says the PyPI names fivetran-mcp and mcp-fivetran belong to unrelated projects, so the server is run from the git URL and has no versioned PyPI package. | | Clients, auth and scoping | The agent acts on behalf of the Airbyte user who signed in, and every tool runs with that user's Airbyte permissions. Direct reads use the credentials stored in the connector rather than the asking user, so source-side access is not per-user. With the context layer enabled, agents can read live source data, query supported destinations and search synced data by keyword, meaning or both. | The server uses one Fivetran API key and secret, so agent actions run with that credential's scope instead of the asking user's identity. The README states OAuth resource-server support in streamable-http mode is not yet functional. Two discovery tools plus one execution tool per allowed resource/action pair keep the tool list small and filtered by scope. | | Access and safety | Read-only mode, enabled with the X-MCP-Readonly-Mode header or ?read-only=true, shows only read-only tools and hides run/cancel sync tools. Permanent deletes only work when the resource name contains delete-me or deleteme. Headers can only hide tools and cannot grant permissions the user's Airbyte role lacks. Direct reads and searches are read-only. | FIVETRAN_SCOPE defaults to read and can be raised to read/write or read/write/delete. DISALLOWED_ACTIONS denies resource:action pairs or single endpoints, and the README advises a denylist for credential endpoints because they outlive the session. Write and delete confirmation is advisory only; the server does not enforce it. | | What it reaches | Doc examples cover Postgres, BigQuery, Snowflake, HubSpot and Salesforce, i.e. ingestion sources and warehouse destinations, but no BI or transformation tools. The hosted server works with all editions of Airbyte Cloud and Airbyte Flex; self-managed Airbyte Core can run it locally with known issues. Clients must support remote MCP servers with OAuth and the Streamable HTTP transport. | Resource groups include connections, destinations, groups, external-logging, external-secrets-managers, hybrid-deployment-agents, connector-sdk and account. The server runs over stdio by default and can run as a hosted HTTP server with --transport streamable-http. | | Lineage and impact | Describe_connection returns a connection's streams, schedule and status, and discover_source_catalog is a catalog tool (stream level, no column lineage documented). None found: Any lineage tool in the Airbyte MCP tools page (column, transformation or BI lineage). | The server exposes every endpoint in the Fivetran API through generated tools; connection schema, table and column configuration endpoints are in the OpenAPI manifest (configuration level, not lineage). None found: Any lineage path in the OpenAPI definition the server loads (searched fivetran-open-api-definition.json for 'lineage' in endpoint paths). Get_schema returns the parameters, request body and response schema of an endpoint, and for connection create/modify it can add service-specific configuration fields. | | Definitions, ownership and rules | Connector skills describe the entities, actions and parameters each connector supports, which is connector documentation rather than business definitions. None found: Glossary, ownership or classification tools in the Airbyte MCP tools page. | The only metadata resource in the action list is metadata:read, which maps to the connector-types endpoints (/v1/metadata/connector-types). None found: Glossary, ownership or classification tools in the Fivetran MCP README or AVAILABLE_ACTIONS.md. | | Freshness and upkeep | Tools check sync status, read job logs and explain failures, so pipeline state is read live from Airbyte. None found: Scheduled or event-driven metadata collection described in the Airbyte MCP docs; tools query on demand. | The README's example questions are about live pipeline state, such as the last time a postgres connection completed a sync and whether any connections are broken. The OpenAPI manifest includes sync-history, state and sync endpoints per connection. | | Profile | Each organization gets 100,000 direct reads and searches per month at no cost; other MCP tools do not count toward the limit. | The README lists a copy-connections plugin for Claude Code and Codex that uses this server, hosted in its own repository. Setup requires Python 3.10+ and uv; API credentials are generated in the Fivetran dashboard. | Each cell joins the facts read for that topic; the full list with sources is on each entry. Read 10 Nov 2026. ## Tool names, as published ### Airbyte MCP (hosted) `get_default_workspace_context`, `list_organizations`, `describe_organization`, `get_organization_billing_status`, `list_workspaces`, `describe_workspace`, `set_default_workspace`, `list_connectors`, `describe_connector`, `deploy_connector`, `deploy_noop_destination`, `check_connector`, `rename_connector`, `update_connector_config`, `permanently_delete_connector`, `list_connections`, `describe_connection`, `create_connection`, `update_connection`, `rename_connection`, `set_connection_selected_streams`, `set_connection_table_prefix`, `discover_source_catalog`, `get_connection_artifact`, `permanently_delete_connection`, `run_connection_sync`, `cancel_connection_sync`, `get_connection_sync_status`, `list_connection_sync_jobs`, `get_connection_sync_logs`, `show_workspace_sync_status`, `show_connection_sync_history`, `show_connector_definitions`, `list_custom_source_definitions`, `get_custom_source_definition`, `get_connector_builder_draft_manifest`, `publish_custom_source_definition`, `update_custom_source_definition`, `permanently_delete_custom_source_definition`, `get_external_access_skill_docs`, `execute_external_api_query`, `execute_external_sql_query`, `execute_external_search_query`, `get_search_index_status`, `request_search_index_backfill`, `list_connector_definitions`, `get_connector_definition`, `get_connector_definition_version_history`, `get_api_docs_urls`, `search_airbyte_knowledge_sources` ### Fivetran MCP Server `list_endpoints`, `get_schema`, `account_read`, `connections_delete`, `connections_read`, `connections_write`, `connector_sdk_delete`, `connector_sdk_read`, `connector_sdk_write`, `destinations_delete`, `destinations_read`, `destinations_write`, `external_logging_delete`, `external_logging_read`, `external_logging_write`, `external_secrets_managers_delete`, `external_secrets_managers_read`, `external_secrets_managers_write`, `external_secrets_managers_entities_read`, `groups_delete`, `groups_read`, `groups_write`, `hvr_write`, `hybrid_deployment_agents_delete`, `hybrid_deployment_agents_read`, `hybrid_deployment_agents_write`, `metadata_read`, `private_links_delete`, `private_links_read`, `private_links_write`, `proxy_delete`, `proxy_read`, `proxy_write`, `public_read`, `roles_read`, `system_keys_delete`, `system_keys_read`, `system_keys_write`, `teams_delete`, `teams_read`, `teams_write`, `transformation_projects_delete`, `transformation_projects_read`, `transformation_projects_write`, `transformations_delete`, `transformations_read`, `transformations_write`, `users_delete`, `users_read`, `users_write`, `webhooks_delete`, `webhooks_read`, `webhooks_write` ## Read next - [Airbyte MCP (hosted), the full entry](/servers/airbyte-mcp/) - [Fivetran MCP Server, the full entry](/servers/fivetran-mcp/) - [Every side-by-side](/compare/) ## Sources - docs.airbyte.com/platform/airbyte-mcp.md: https://docs.airbyte.com/platform/airbyte-mcp.md (read 10 Nov 2026) - docs.airbyte.com/platform/airbyte-mcp/install.md: https://docs.airbyte.com/platform/airbyte-mcp/install.md (read 10 Nov 2026) - pypi.org/pypi/airbyte/json: https://pypi.org/pypi/airbyte/json (read 10 Nov 2026) - docs.airbyte.com/llms.txt: https://docs.airbyte.com/llms.txt (read 10 Nov 2026) - docs.airbyte.com/platform/airbyte-mcp/tools.md: https://docs.airbyte.com/platform/airbyte-mcp/tools.md (read 10 Nov 2026) - docs.airbyte.com/platform/airbyte-mcp/advanced-configuration.md: https://docs.airbyte.com/platform/airbyte-mcp/advanced-configuration.md (read 10 Nov 2026) - raw.githubusercontent.com/fivetran/fivetran-mcp/main/README.md: https://raw.githubusercontent.com/fivetran/fivetran-mcp/main/README.md (read 10 Nov 2026) - github.com/fivetran/fivetran-mcp/releases: https://github.com/fivetran/fivetran-mcp/releases (read 10 Nov 2026) - raw.githubusercontent.com/fivetran/fivetran-mcp/main/fivetran-open-api-definition.json: https://raw.githubusercontent.com/fivetran/fivetran-mcp/main/fivetran-open-api-definition.json (read 10 Nov 2026) - raw.githubusercontent.com/fivetran/fivetran-mcp/main/open-api-definitions/AVAILABLE_ACTIONS.md: https://raw.githubusercontent.com/fivetran/fivetran-mcp/main/open-api-definitions/AVAILABLE_ACTIONS.md (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/compare/airbyte-mcp-vs-fivetran-mcp/ --- # Sigma MCP server and ThoughtSpot MCP Server, field by field: 0 and 20 tools named SIDE BY SIDE, read 10 Nov 2026 **The short answer.** Sigma MCP server (Sigma Computing, remote (hosted)) names 0 tools; ThoughtSpot MCP Server (ThoughtSpot, remote (hosted)) names 20. Versions read on 10 Nov 2026: no version number published and 0.5.0. The page puts the same fields in the same rows and draws no conclusion. | Field | Sigma MCP server | ThoughtSpot MCP Server | | --- | --- | --- | | Data tool | [Sigma](/mcp/sigma/) | [ThoughtSpot](/mcp/thoughtspot/) | | Publisher | Sigma Computing | ThoughtSpot | | Kind | Official | Official | | Repository or docs | [help.sigmacomputing.com/docs/use-sigma-mcp-server](https://help.sigmacomputing.com/docs/use-sigma-mcp-server) | [github.com/thoughtspot/mcp-server](https://github.com/thoughtspot/mcp-server) | | Version or commit | Hosted service (no versioned release) | 0.5.0 (npm and package.json); main at commit 820bf16f6aa3 | | Last release | none found on 10 Nov 2026 | 6 Jun 2025 | | Transport | Remote (hosted) | Remote (hosted) | | Auth | OAuth (HTTP transport) inheriting the user's Sigma account permissions; API-key access token with scope mcp:access (Beta). | OAuth to a ThoughtSpot instance (dynamic client registration, or manually registered client id and secret); bearer-token and /sse paths also exist; the user chooses or is bound to a ThoughtSpot instance URL. | | Access | Read and write | Read and write | | Tools listed | 0 named in the evidence | 20 named in the evidence | | Tool definitions read at | [help.sigmacomputing.com/docs/use-sigma-mcp-server.md](https://help.sigmacomputing.com/docs/use-sigma-mcp-server.md) | [raw.githubusercontent.com/thoughtspot/mcp-server/main/src/servers/tool-definitions.ts](https://raw.githubusercontent.com/thoughtspot/mcp-server/main/src/servers/tool-definitions.ts) | | Context weight | Not measured: The docs page describes features (search, explore, analyze, build workbooks, screenshots, agents) in prose with example prompts but publishes no tool names or per-tool list. | at least 2,149 tokens for 20 tool definitions (20 with a description), measured from the source code read | | Status | Exists on 10 Nov 2026 as a Sigma-hosted MCP server (URL per organization under Profile > Integrations). The docs describe capabilities (search, describe, query, build workbooks, screenshots, call agents) but publish no tool names, so tools is empty. | Exists on 10 Nov 2026 as thoughtspot/mcp-server, run as a Cloudflare Worker (hosted at agent.thoughtspot.app) with a local stdio entry in package.json. The GitHub repo has no releases or tags; the npm package last published 6 Jun 2025 although main has moved on, so last_release comes from npm. | | Read | 10 Nov 2026 | 10 Nov 2026 | The same fourteen fields for both servers, read 10 Nov 2026. ## In one sentence each - [Sigma MCP server](/servers/sigma-mcp/): Sigma's hosted server searches data models and workbooks, reads column definitions, and queries and builds workbooks under the user's own Sigma permissions. - [ThoughtSpot MCP Server](/servers/thoughtspot-mcp-server/): ThoughtSpot's server returns Spotter answers, searches objects with owner, tag and verified filters, and creates dashboards and models within the user's privileges. ## The read facts, topic by topic | Topic | Sigma MCP server | ThoughtSpot MCP Server | | --- | --- | --- | | Publisher and releases | Sigma documents its own Sigma MCP server for searching, describing and querying data and building workbooks. The server is a hosted endpoint whose URL is copied from Profile > Integrations; no versioned release or package is published. The API-key access path for the MCP server is labeled Beta. | ThoughtSpot publishes the Spotter MCP Server in its own GitHub organization. The GitHub releases page for thoughtspot/mcp-server shows no releases. The npm package @thoughtspot/mcp-server has latest 0.5.0, published 6 Jun 2025, under a ThoughtSpot end user license agreement; repo package.json also reads 0.5.0. | | Clients, auth and scoping | With OAuth the server inherits the user's Sigma account permissions. Access needs the 'Use Sigma MCP with OAuth' account-type permission plus at least Can view on documents or Can use on connections. Agents can call Sigma agents defined in workbooks if the user has Can view on the workbook. | Spotter 'handles authentication securely, and returns live answers from your own data' per the README. The model tools are offered as a group and gated on the user's data-modeling privileges, and get_data is gated on the data-download privilege at runtime. OAuth client registration can be tied to a ThoughtSpot instance so users do not enter the instance URL. | | Access and safety | Workbooks cannot be built when authenticating with an API key, and builds cannot include AI columns or custom SQL elements. MCP queries can be monitored in the data platform using the "kind":"mcp" query tag. Orgs with an IP allowlist must allowlist the AI assistant's outbound IPs with scope API. | Search_objects and get_data are annotated readOnlyHint true and destructiveHint false. Write tools exist: createLiveboard, create_dashboard, modify_dashboard and the model tools create or change content. INSTANCE_ALLOWLIST_BY_HOST restricts which ThoughtSpot instances each Worker domain accepts, returning 403 for others. | | What it reaches | Documented clients include Claude, ChatGPT, Codex, Cortex Code and Cursor over HTTP transport. The configuration table sets transport to HTTP. | The tool set includes connectivity and auth checks (ping, check_connectivity) and org handling tools list_orgs and switch_org, so one server can reach multiple ThoughtSpot Orgs. The README lists /mcp, /sse, /token/* and /bearer/* endpoints on the Worker. | | Lineage and impact | Agents can explore data sources and documents by asking about column names, definitions, descriptions, data types and column attributes. Search covers documents and data sources including data models, warehouse tables, and elements from workbooks and data models. None found: Lineage or upstream-dependency capability on the Sigma MCP server page. | Search_objects finds Answers, Liveboards and Worksheets with filters for types, owner, tag, modified_since and verified_only, returning object metadata (title, description, query tokens, tags). The tool source warns that description and query tokens are authoring metadata that can be stale, and advises grounding explanations in the object's data via get_data. None found: A lineage tool in tool-definitions.ts (no column, upstream-table or transformation lineage tool among the 20 tools). | | Definitions, ownership and rules | The server tells the assistant about existing filters and controls that affect a workbook answer so it can disclose or override them. | Search_objects results include a Verified flag and Owner, and tags, so certification and ownership signals reach the agent. A V3 'Spotter Model' tool group (create_model_session, send_model_message, get_model_updates, finalize_model) lets an agent create and edit data models. | | Freshness and upkeep | The server queries only the published version of a document, not custom views, saved views or tagged versions. None found: Scheduled or event-driven metadata collection on the Sigma MCP page; queries run at call time. | Get_data re-runs a saved Answer or Liveboard against current data, so rows reflect the latest data. None found: Scheduled or event-driven metadata collection in the ThoughtSpot MCP README; tools query the instance per request. | | Profile | New workbooks built over MCP are saved in the user's Your documents folder; editing an existing workbook publishes a new version. | The README links a free ThoughtSpot trial and a hosted client registration page at agent.thoughtspot.app. | Each cell joins the facts read for that topic; the full list with sources is on each entry. Read 10 Nov 2026. ## Tool names, as published ### Sigma MCP server No tool names are published in the source read on 10 Nov 2026. ### ThoughtSpot MCP Server `ping`, `getRelevantQuestions`, `getAnswer`, `createLiveboard`, `getDataSourceSuggestions`, `check_connectivity`, `search_objects`, `get_data`, `create_analysis_session`, `send_session_message`, `get_session_updates`, `create_dashboard`, `modify_dashboard`, `get_dashboard_status`, `list_orgs`, `switch_org`, `create_model_session`, `send_model_message`, `get_model_updates`, `finalize_model` ## Read next - [Sigma MCP server, the full entry](/servers/sigma-mcp/) - [ThoughtSpot MCP Server, the full entry](/servers/thoughtspot-mcp-server/) - [Every side-by-side](/compare/) ## Sources - help.sigmacomputing.com/docs/use-sigma-mcp-server: https://help.sigmacomputing.com/docs/use-sigma-mcp-server (read 10 Nov 2026) - github.com/thoughtspot/mcp-server: https://github.com/thoughtspot/mcp-server (read 10 Nov 2026) - github.com/thoughtspot/mcp-server/releases: https://github.com/thoughtspot/mcp-server/releases (read 10 Nov 2026) - registry.npmjs.org/@thoughtspot/mcp-server: https://registry.npmjs.org/@thoughtspot/mcp-server (read 10 Nov 2026) - raw.githubusercontent.com/thoughtspot/mcp-server/main/src/servers/tool-definitions.ts: https://raw.githubusercontent.com/thoughtspot/mcp-server/main/src/servers/tool-definitions.ts (read 10 Nov 2026) - raw.githubusercontent.com/thoughtspot/mcp-server/main/README.md: https://raw.githubusercontent.com/thoughtspot/mcp-server/main/README.md (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/compare/sigma-mcp-vs-thoughtspot-mcp-server/ --- # Hex MCP server and Omni MCP Server, field by field: 18 and 26 tools named SIDE BY SIDE, read 10 Nov 2026 **The short answer.** Hex MCP server (Hex, remote (hosted)) names 18 tools; Omni MCP Server (Omni Analytics, remote (hosted)) names 26. Versions read on 10 Nov 2026: no version number published and no version number published. The page puts the same fields in the same rows and draws no conclusion. | Field | Hex MCP server | Omni MCP Server | | --- | --- | --- | | Data tool | [Hex](/mcp/hex/) | [Omni](/mcp/omni/) | | Publisher | Hex | Omni Analytics | | Kind | Official | Official | | Repository or docs | [learn.hex.tech/docs/api-integrations/mcp-server](https://learn.hex.tech/docs/api-integrations/mcp-server) | [docs.omni.co/ai/mcp](https://docs.omni.co/ai/mcp) | | Version or commit | Hosted service (no versioned release) | Hosted service (no versioned release) | | Last release | none found on 10 Nov 2026 | none found on 10 Nov 2026 | | Transport | Remote (hosted) | Remote (hosted) | | Auth | OAuth authorization to the user's Hex workspace (server URL app.hex.tech/mcp, or the customer's single-tenant, EU or HIPAA domain); requires Explorer role or higher. | OAuth 2.1 with PKCE (creates an MCP OAuth PAT for the user), or an API key (personal token or Organization API key; Organization keys can act as a user with X-MCP-User-ID). | | Access | Read and write | Configurable | | Tools listed | 18 named in the evidence | 26 named in the evidence | | Tool definitions read at | [learn.hex.tech/docs/api-integrations/mcp-server](https://learn.hex.tech/docs/api-integrations/mcp-server) | [docs.omni.co/ai/mcp/tools.md](https://docs.omni.co/ai/mcp/tools.md) | | Context weight | 252 tokens for 18 tool definitions (18 with a description), measured from the docs page read | 466 tokens for 27 tool definitions (27 with a description), measured from the docs page read | | Status | Exists on 10 Nov 2026 as a Hex-hosted MCP server; tool list read from the Hex docs page (knowledge tools for Explorers, project editing tools for Editors). | Exists on 10 Nov 2026 as Omni's hosted MCP server (OAuth callback via callbacks.omniapp.co); admins toggle tool capabilities in AI Hub > MCP. The docs page lists 25 tools. | | Read | 10 Nov 2026 | 10 Nov 2026 | The same fourteen fields for both servers, read 10 Nov 2026. ## In one sentence each - [Hex MCP server](/servers/hex-mcp/): Hex's hosted server runs Threads (agent analyses over the user's data connections) and notebook editing under workspace roles and project permissions. - [Omni MCP Server](/servers/omni-mcp/): Omni's hosted server runs as a specific Omni user and exposes models, topics, saved queries and apps, so answers follow the semantic model and the user's permissions. ## The read facts, topic by topic | Topic | Hex MCP server | Omni MCP Server | | --- | --- | --- | | Publisher and releases | Hex documents its own MCP server and an official Hex plugin in the Cursor Marketplace. No versioned release is published; the hosted URL is app.hex.tech/mcp. | Omni documents its own MCP server for querying Omni data and searching Omni docs from external AI tools. No versioned release is published for the hosted server; it is enabled by an organization setting (AI Hub > MCP, enabled by default). | | Clients, auth and scoping | Threads created through the MCP server behave like Threads in the Hex app and select among the data connections the user can access. The MCP server enforces workspace roles and project permissions; editing a cell needs Can Edit or Full Access on that project, even for Admins. Knowledge tools need the Explorer role or above and project editing tools need Editor or above. | MCP tools run with the permissions of a specific Omni user, so results include only what that user can access. With OAuth, all MCP queries use the user's Omni permissions and access controls. An Organization API key can run tools as another user via the X-MCP-User-ID header, and X-MCP-Model-ID can pin a model. | | Access and safety | The MCP server does not query Sensitive data connections by default and cannot read cell outputs from projects that use them, unless an admin allows it. MCP access is not governed by the workspace API access setting; deactivating the user is how to revoke all of a user's access. The tools include create, update and delete for cells, so the server is read-write for Editors. | The tool list includes app write tools (putApp, patchApp, publishDraft) so the server is not read-only; edits land on a shared draft until publishDraft runs. Organization Admins can individually enable or disable tool capabilities on the MCP settings page. A user-scoped key lists a tool only if the member holds the permission it needs. QueryUsageAnalytics uses read-only queries on usage data and is limited to Organization Admins. | | What it reaches | The server works with Claude Desktop, Claude Code, Cursor, ChatGPT, Codex and Glean as well as standard MCP clients. Single-tenant, EU multi-tenant and HIPAA multi-tenant workspaces use their own host instead of app.hex.tech. | The same MCP server can serve embed users, scoped to each user's permissions. Setup pages exist for ChatGPT, Claude Code, Claude Desktop, Codex, Cursor, Microsoft Copilot and VS Code. | | Lineage and impact | Project editing tools list cells and read cell outputs and chart images, giving agents the SQL, Python and Markdown cells of a notebook. Search_projects finds projects in the workspace, and get_project reads a project's metadata. None found: A lineage tool on the Hex MCP server page (cell-level reads exist, but no column or upstream-table lineage tool is listed). | PickModel returns the models available to the calling user and pickTopic returns the topics in the chosen model and selects the most relevant one. SearchContent finds existing dashboards and apps by name, folder or tile name, and listDocumentQueries lists the saved queries in a document. None found: A lineage tool among the Omni MCP tools (no column or upstream-table lineage tool documented). | | Definitions, ownership and rules | Admins can mark data connections Sensitive and the docs point to descriptions and exclusions as the way to shape what the Hex Agent sees. | Queries run against Omni's modeled topics, and the Omni docs publish guidance on configuring AI context at model, topic and view level. ListSkills finds a saved Omni skill by name, identifier or description and reads its instructions without running it. | | Freshness and upkeep | Run_cell and run_notebook execute cells or the draft notebook and get_run polls status, so results are produced at call time. None found: Scheduled metadata collection on the Hex MCP page. | None found: Scheduled or event-driven metadata collection in the Omni MCP docs; tools run queries and read content at call time. GetData and runQuery execute queries on request, and checkStatus and cancelJob manage running jobs. | | Profile | The llms.txt notes workspace data is private and MCP access requires the user to authorize an OAuth connection. Claude connector widgets created by the Hex connector expire after 24 hours, and file uploads to a conversation are not supported. | The OAuth connection URL routes requests to the user's Omni organization through callbacks.omniapp.co/callback/mcp. | Each cell joins the facts read for that topic; the full list with sources is on each entry. Read 10 Nov 2026. ## Tool names, as published ### Hex MCP server `search_projects`, `create_thread`, `get_thread`, `continue_thread`, `get_me`, `create_project`, `get_project`, `list_cells`, `get_cell`, `create_cell`, `update_cell`, `delete_cell`, `run_cell`, `run_notebook`, `get_run`, `get_cell_output`, `get_cell_image`, `list_data_connections` ### Omni MCP Server `pickModel`, `pickTopic`, `getData`, `runQuery`, `listLlms`, `askOmni`, `checkStatus`, `cancelJob`, `listBranches`, `listSkills`, `submitFeedback`, `searchContent`, `listDocumentQueries`, `runSavedQuery`, `startDashboardDownload`, `checkDashboardDownload`, `fetchDashboardDownload`, `getApp`, `getAppDraft`, `putApp`, `patchApp`, `publishDraft`, `showApp`, `import-claude-dashboard-from-url`, `get-dashboard-import-job-status`, `queryUsageAnalytics` ## Read next - [Hex MCP server, the full entry](/servers/hex-mcp/) - [Omni MCP Server, the full entry](/servers/omni-mcp/) - [Every side-by-side](/compare/) ## Sources - learn.hex.tech/docs/api-integrations/mcp-server: https://learn.hex.tech/docs/api-integrations/mcp-server (read 10 Nov 2026) - learn.hex.tech/llms.txt: https://learn.hex.tech/llms.txt (read 10 Nov 2026) - docs.omni.co/ai/mcp/tools.md: https://docs.omni.co/ai/mcp/tools.md (read 10 Nov 2026) - docs.omni.co/ai/mcp/authentication.md: https://docs.omni.co/ai/mcp/authentication.md (read 10 Nov 2026) - docs.omni.co/llms.txt: https://docs.omni.co/llms.txt (read 10 Nov 2026) - docs.omni.co/sitemap.xml: https://docs.omni.co/sitemap.xml (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/compare/hex-mcp-vs-omni-mcp/ --- # Atlan MCP Server and Euno MCP, field by field: 41 and 56 tools named SIDE BY SIDE, read 10 Nov 2026 **The short answer.** Atlan MCP Server (Atlan, remote (hosted)) names 41 tools; Euno MCP (Euno, remote (hosted)) names 56. Versions read on 10 Nov 2026: 0.3.3 and no version number published. The page puts the same fields in the same rows and draws no conclusion. | Field | Atlan MCP Server | Euno MCP | | --- | --- | --- | | Data tool | [Atlan](/mcp/atlan/) | [Euno](/mcp/euno/) | | Publisher | Atlan | Euno | | Kind | Official | Official | | Repository or docs | [docs.atlan.com/product/capabilities/atlan-ai/references/mcp-tools](https://docs.atlan.com/product/capabilities/atlan-ai/references/mcp-tools) | [docs.euno.ai/ai/mcp-setup](https://docs.euno.ai/ai/mcp-setup?utm_source=datamcpservers.com&utm_medium=referral&utm_campaign=side-by-side) | | Version or commit | Hosted at mcp.atlan.com/mcp (unversioned); local atlan-mcp-server 0.3.3 (deprecated) | Hosted service at api.app.euno.ai/mcp, no version published (docs read 10 Nov 2026) | | Last release | 17 Feb 2026 | none found on 10 Nov 2026 | | Transport | Remote (hosted) | Remote (hosted) | | Auth | OAuth per user (calls run as the signed-in user) or Atlan API key as Bearer token (shared service identity); local server uses API key + base URL. | OAuth sign-in (default) or API key and headers; active persona scopes data and actions. | | Access | Configurable | Configurable | | Tools listed | 41 named in the evidence | 56 named in the evidence | | Tool definitions read at | [docs.atlan.com/product/capabilities/atlan-ai/references/mcp-tools](https://docs.atlan.com/product/capabilities/atlan-ai/references/mcp-tools) | [docs.euno.ai/ai/mcp-setup.md](https://docs.euno.ai/ai/mcp-setup.md?utm_source=datamcpservers.com&utm_medium=referral&utm_campaign=side-by-side) | | Context weight | 1,770 tokens for 42 tool definitions (42 with a description), measured from the docs page read | 760 tokens for 56 tool definitions (42 with a description), measured from the docs page read | | Status | Exists on 10 Nov 2026: hosted Atlan MCP (41 tools, page updated 27 Oct 2026); the local atlan-mcp-server in atlanhq/agent-toolkit is marked deprecated in its README. | Exists on 10 Nov 2026; documented in docs.euno.ai MCP Setup Guide; tool names and descriptions published, parameter schemas not published. | | Read | 10 Nov 2026 | 10 Nov 2026 | The same fourteen fields for both servers, read 10 Nov 2026. ## In one sentence each - [Atlan MCP Server](/servers/atlan-mcp/): Atlan's hosted per-tenant server runs as the signed-in user, scopes context by persona, offers an optional read-only mode and proposes writes for approval. - [Euno MCP](/servers/euno-mcp/): Euno's hosted OAuth endpoint serves persona-scoped lineage, impact analysis, EQL search and catalog tools over its metadata graph, with no warehouse query tool. ## The read facts, topic by topic | Topic | Atlan MCP Server | Euno MCP | | --- | --- | --- | | Publisher and releases | Atlan hosts the MCP server itself at mcp.atlan.com/mcp, per tenant. The older local server in atlanhq/agent-toolkit is deprecated in favor of the hosted endpoint. The last GitHub release of the local server is v0.3.3, on 17 Feb 2026; the hosted server has no public version number. | Euno publishes and documents its own MCP server at api.app.euno.ai/mcp. None found: A version number or dated release list for the Euno MCP server; the docs describe a hosted service, and the separate Euno SDK is v0.5. Stdio use goes through the third-party open-source mcp-remote bridge, not a Euno-published package. | | Clients, auth and scoping | The MCP server serves the catalog as a context layer to AI clients (Claude, Cursor, ChatGPT, Gemini, Copilot) and automation platforms. OAuth makes each tool call run as the signed-in user, so tools run with their own permissions and actions are attributed to them. Guests and members are scoped to their persona; MCP narrows but never widens what a user's login permits. The tools page lists 41 tools across nine categories. | Euno MCP uses the active persona; if none can be identified the user is asked to choose. Euno MCP works with Cursor, Claude, ChatGPT in Developer Mode and Claude Code. Euno MCP has a dedicated Gemini Enterprise setup path besides direct, bridge and API-key setups. | | Access and safety | An optional read-only mode removes every write, admin and lifecycle tool while keeping search and lineage; it is requested from Atlan support. Tools that change anything propose first and wait for approval before writing. Every tool call is logged as a structured event with tool name, client, tenant, request ID and status. Tools that execute SQL route it through a read-only validator that permits only SELECT-class queries. | Euno MCP includes read and write tools; write tools such as catalog creates and benchmark changes ask for client confirmation. Euno MCP catalog reads are published read-only and catalog writes are marked destructive; deletes need confirm=true. Euno enforces authentication and persona permissions on the server, independent of client confirmation. Context benchmark tools are accepted only for an account admin persona. | | What it reaches | The docs list supported AI clients: Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, ChatGPT, Snowflake, Google ADK, n8n, Microsoft 365 Copilot, Glean and Databricks. Data-source coverage depends on Atlan connectors, not stated on this page. | The published tool table covers metadata, lineage, EQL, audit, documentation, catalog and benchmark tools; no listed tool executes SQL against a warehouse, and query_to_eql converts a filter without executing it. Euno's FAQ says Euno does not query customer data. | | Lineage and impact | The hosted Atlan MCP exposes a traverse_lineage tool for upstream/downstream lineage with deeper hops on request. The docs say the hosted Atlan MCP can traverse end-to-end lineage; the tool page does not state column-level or SQL-logic parsing for the lineage tool. None found: A statement on the MCP tools page that traverse_lineage returns column-level edges or transformation SQL. Query_assets runs read-only SQL against a connected warehouse (Snowflake, Redshift, BigQuery, Databricks), returning up to 100 rows. | Euno MCP exposes get_upstream_lineage and resource_impact_analysis tools for one resource URI. Euno MCP can search and aggregate the metadata graph with EQL (search, count, multi-count, pivot). Euno MCP's example prompt traces a Tableau dashboard's upstream lineage. | | Definitions, ownership and rules | Tools cover glossaries, glossary categories and terms, domains and data products, custom metadata sets, and certification/descriptions. Add_atlan_tags applies classification tags with propagation to downstream lineage on by default. Resolve_metadata returns tenant-level tag and custom-metadata definitions reconciled live, including enum option values. The lineage tool returns owners, certificate status, tags and readme with each asset by default (local server docstring). | Euno MCP has glossary, metadata tag and workflow tools (14 direct catalog tools). Euno MCP's load_account_settings returns account, persona, skill and metadata context; skills carry organization-specific workflows. Euno MCP includes sensitive_data_incident_response returning evidence per user and BI-resource pair. | | Freshness and upkeep | Metadata is retrieved in real time with no intermediate caching. | The Euno MCP server keeps no session state, so connections survive restarts and scaling. Euno's October 2026 release notes add pivot and multi-count aggregation tools through MCP. | | Profile | Atlan's compliance posture for MCP is covered by HIPAA, GDPR and CCPA/CPRA programs, with certifications on the Atlan Trust Portal. | Euno's hosted MCP uses OAuth; an API-key setup exists for clients that cannot do interactive sign-in. | Each cell joins the facts read for that topic; the full list with sources is on each entry. Read 10 Nov 2026. ## Tool names, as published ### Atlan MCP Server `semantic_search`, `ask`, `fluent_search`, `fluent_search_schema`, `search_assets`, `get_assets`, `search_atlan_docs`, `resolve_metadata`, `get_groups`, `get_users`, `describe_asset_type`, `traverse_lineage`, `query_assets`, `update_assets`, `update_custom_metadata`, `add_atlan_tags`, `remove_atlan_tag`, `manage_announcements`, `create_custom_metadata_set`, `add_attributes_to_cm_set`, `remove_attributes_from_cm_set`, `remove_custom_metadata`, `delete_custom_metadata_set`, `manage_asset_lifecycle`, `create_glossaries`, `create_glossary_categories`, `create_glossary_terms`, `create_domains`, `create_data_products`, `create_dq_rules`, `update_dq_rules`, `schedule_dq_rules`, `delete_dq_rules`, `read_artifact`, `search_knowledge`, `list_knowledge`, `read_knowledge_file`, `create_knowledge_folder`, `save_knowledge_content`, `get_knowledge_upload_url`, `register_knowledge_file` ### Euno MCP `load_account_settings`, `list_personas`, `current_persona`, `switch_persona`, `load_skill`, `search_properties`, `fetch_single_resource`, `find_resource_by_name`, `find_resources_for_topic`, `get_upstream_lineage`, `resource_impact_analysis`, `query_to_eql`, `search_resources_eql`, `count_resources_eql`, `multi_count_resources_eql`, `pivot_resources_eql`, `search_resource_audit_log`, `sensitive_data_incident_response`, `documentation_search`, `documentation_get_full_document`, `documentation_get_surrounding_context`, `list_metadata_tags`, `list_metadata_tag_domains`, `preview_metadata_tag_plan`, `get_context_benchmark_overview`, `get_context_benchmark_questions`, `list_context_benchmark_runs`, `get_context_benchmark_run`, `get_context_benchmark_improvement_report`, `get_context_benchmark_question_result`, `get_context_benchmark_suggestion_application`, `get_context_benchmark_schedule`, `create_context_benchmark_question`, `update_context_benchmark_question`, `delete_context_benchmark_question`, `start_context_benchmark_run`, `stop_context_benchmark_run`, `update_context_benchmark_schedule`, `prepare_context_benchmark_suggestion_application`, `decide_context_benchmark_suggestion_application`, `undo_context_benchmark_suggestion`, `reapply_context_benchmark_suggestion`, `fetch_glossary_term`, `find_glossary_term`, `create_glossary_term`, `update_glossary_term`, `delete_glossary_term`, `fetch_metadata_tag`, `create_metadata_tag`, `update_metadata_tag`, `delete_metadata_tag`, `fetch_workflow`, `find_workflow`, `create_workflow`, `update_workflow`, `delete_workflow` **What Euno's server does not do.** Euno's server gives an agent governed context about the data; it does not run the query, which is the job of the warehouse's own server. ## Read next - [Atlan MCP Server, the full entry](/servers/atlan-mcp/) - [Euno MCP, the full entry](/servers/euno-mcp/) - [Every side-by-side](/compare/) ## Sources - docs.atlan.com/product/capabilities/atlan-ai/references/mcp-tools: https://docs.atlan.com/product/capabilities/atlan-ai/references/mcp-tools (read 10 Nov 2026) - docs.atlan.com/product/capabilities/atlan-ai/how-tos/remote-mcp-overview: https://docs.atlan.com/product/capabilities/atlan-ai/how-tos/remote-mcp-overview (read 10 Nov 2026) - github.com/atlanhq/agent-toolkit/blob/main/modelcontextprotocol/server.py: https://github.com/atlanhq/agent-toolkit/blob/main/modelcontextprotocol/server.py (read 10 Nov 2026) - docs.atlan.com/product/capabilities/atlan-ai/references/mcp-security: https://docs.atlan.com/product/capabilities/atlan-ai/references/mcp-security (read 10 Nov 2026) - github.com/atlanhq/agent-toolkit/blob/main/modelcontextprotocol/README.md: https://github.com/atlanhq/agent-toolkit/blob/main/modelcontextprotocol/README.md (read 10 Nov 2026) - github.com/atlanhq/agent-toolkit/blob/main/CHANGELOG.md: https://github.com/atlanhq/agent-toolkit/blob/main/CHANGELOG.md (read 10 Nov 2026) - docs.euno.ai/ai/mcp-setup.md: https://docs.euno.ai/ai/mcp-setup.md (read 10 Nov 2026) - euno.ai: https://euno.ai/ (read 10 Nov 2026) - docs.euno.ai/support-resources/release-note/october-2026.md: https://docs.euno.ai/support-resources/release-note/october-2026.md (read 10 Nov 2026) - docs.euno.ai/ai/mcp-setup/mcp-bridge-setup.md: https://docs.euno.ai/ai/mcp-setup/mcp-bridge-setup.md (read 10 Nov 2026) - docs.euno.ai/ai/mcp-api-key-setup.md: https://docs.euno.ai/ai/mcp-api-key-setup.md (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/compare/atlan-mcp-vs-euno-mcp/ --- # DataHub MCP Server and Euno MCP, field by field: 22 and 56 tools named SIDE BY SIDE, read 10 Nov 2026 **The short answer.** DataHub MCP Server (Acryl Data, remote and local) names 22 tools; Euno MCP (Euno, remote (hosted)) names 56. Versions read on 10 Nov 2026: 0.8.0 and no version number published. The page puts the same fields in the same rows and draws no conclusion. | Field | DataHub MCP Server | Euno MCP | | --- | --- | --- | | Data tool | [DataHub](/mcp/datahub/) | [Euno](/mcp/euno/) | | Publisher | Acryl Data (DataHub) | Euno | | Kind | Official | Official | | Repository or docs | [github.com/acryldata/mcp-server-datahub](https://github.com/acryldata/mcp-server-datahub) | [docs.euno.ai/ai/mcp-setup](https://docs.euno.ai/ai/mcp-setup?utm_source=datamcpservers.com&utm_medium=referral&utm_campaign=side-by-side) | | Version or commit | 0.8.0 | Hosted service at api.app.euno.ai/mcp, no version published (docs read 10 Nov 2026) | | Last release | 21 Oct 2026 | none found on 10 Nov 2026 | | Transport | Remote and local | Remote (hosted) | | Auth | Stdio: DATAHUB_GMS_URL + DATAHUB_GMS_TOKEN (or ~/.datahubenv); HTTP: per-request DataHub personal access token as Authorization Bearer, no shared server token. | OAuth sign-in (default) or API key and headers; active persona scopes data and actions. | | Access | Configurable | Configurable | | Tools listed | 22 named in the evidence | 56 named in the evidence | | Tool definitions read at | [github.com/acryldata/mcp-server-datahub/blob/main/README.md](https://github.com/acryldata/mcp-server-datahub/blob/main/README.md) | [docs.euno.ai/ai/mcp-setup.md](https://docs.euno.ai/ai/mcp-setup.md?utm_source=datamcpservers.com&utm_medium=referral&utm_campaign=side-by-side) | | Context weight | 567 tokens for 22 tool definitions (22 with a description), measured from the README read | 760 tokens for 56 tool definitions (42 with a description), measured from the docs page read | | Status | Exists under this name on 10 Nov 2026; v0.8.0 released 21 Oct 2026 per GitHub releases and PyPI. | Exists on 10 Nov 2026; documented in docs.euno.ai MCP Setup Guide; tool names and descriptions published, parameter schemas not published. | | Read | 10 Nov 2026 | 10 Nov 2026 | The same fourteen fields for both servers, read 10 Nov 2026. ## In one sentence each - [DataHub MCP Server](/servers/datahub-mcp/): DataHub's server returns column-level lineage with paths and SQL transformations, plus glossary, domain and owner discovery, with per-user tokens in HTTP mode. - [Euno MCP](/servers/euno-mcp/): Euno's hosted OAuth endpoint serves persona-scoped lineage, impact analysis, EQL search and catalog tools over its metadata graph, with no warehouse query tool. ## The read facts, topic by topic | Topic | DataHub MCP Server | Euno MCP | | --- | --- | --- | | Publisher and releases | mcp-server-datahub is published from the acryldata GitHub organization (DataHub's company) and is linked from DataHub's own docs at docs.datahub.com. The GitHub release page shows v0.8.0 as a published release (PyPI upload 21 Oct 2026). The package is on PyPI as mcp-server-datahub 0.8.0. | Euno publishes and documents its own MCP server at api.app.euno.ai/mcp. None found: A version number or dated release list for the Euno MCP server; the docs describe a hosted service, and the separate Euno SDK is v0.5. Stdio use goes through the third-party open-source mcp-remote bridge, not a Euno-published package. | | Clients, auth and scoping | The server is built to serve DataHub context to AI agents: search, lineage, query history and entity metadata. In the Docker HTTP deployment each MCP client supplies its own DataHub token, so DataHub permissions and audit identity are preserved per user. An optional get_me tool (TOOLS_IS_USER_ENABLED=true) returns the authenticated user's profile and group memberships. | Euno MCP uses the active persona; if none can be identified the user is asked to choose. Euno MCP works with Cursor, Claude, ChatGPT in Developer Mode and Claude Code. Euno MCP has a dedicated Gemini Enterprise setup path besides direct, bridge and API-key setups. | | Access and safety | Mutation tools are disabled by default and need TOOLS_IS_MUTATION_ENABLED=true, so the default is read-only. The HTTP entry point refuses to start if a shared server-wide token is configured, forcing per-user tokens. Tokens are accepted only in the Authorization header; query-string tokens are rejected. | Euno MCP includes read and write tools; write tools such as catalog creates and benchmark changes ask for client confirmation. Euno MCP catalog reads are published read-only and catalog writes are marked destructive; deletes need confirm=true. Euno enforces authentication and persona permissions on the server, independent of client confirmation. Context benchmark tools are accepted only for an account admin persona. | | What it reaches | The README says the server shows data in databases, lakes, warehouses and BI tools registered in DataHub; no connector list is on this page. | The published tool table covers metadata, lineage, EQL, audit, documentation, catalog and benchmark tools; no listed tool executes SQL against a warehouse, and query_to_eql converts a filter without executing it. Euno's FAQ says Euno does not query customer data. | | Lineage and impact | DataHub's MCP server exposes a get_lineage tool covering datasets, columns and dashboards with hop control. The README says lineage is traceable at table and column level and that transformation logic is understood. A get_lineage_paths_between tool returns exact lineage paths between two assets or columns, including intermediate transformations and SQL query information. Get_dataset_queries fetches real SQL queries that reference a dataset or column. | Euno MCP exposes get_upstream_lineage and resource_impact_analysis tools for one resource URI. Euno MCP can search and aggregate the metadata graph with EQL (search, count, multi-count, pivot). Euno MCP's example prompt traces a Tableau dashboard's upstream lineage. | | Definitions, ownership and rules | DataHub's MCP server lets agents discover domains, owners, tags and glossary terms. Optional mutation tools (off by default) add glossary terms, tags, owners, domains, descriptions and structured properties; this is metadata writing, and the docs do not say terms propagate along lineage through MCP. None found: Any statement in the DataHub MCP README that tags, terms or classifications propagate along lineage when read through the MCP tools. | Euno MCP has glossary, metadata tag and workflow tools (14 direct catalog tools). Euno MCP's load_account_settings returns account, persona, skill and metadata context; skills carry organization-specific workflows. Euno MCP includes sensitive_data_incident_response returning evidence per user and BI-resource pair. | | Freshness and upkeep | None found: Any freshness or ingestion-cadence statement in the DataHub MCP README (the server reads whatever DataHub has ingested). The README only describes caching of verified tokens for five minutes. | The Euno MCP server keeps no session state, so connections survive restarts and scaling. Euno's October 2026 release notes add pivot and multi-count aggregation tools through MCP. | | Profile | DataHub describes itself as an open-source context platform; the MCP server is Apache-2.0 licensed Python. | Euno's hosted MCP uses OAuth; an API-key setup exists for clients that cannot do interactive sign-in. | Each cell joins the facts read for that topic; the full list with sources is on each entry. Read 10 Nov 2026. ## Tool names, as published ### DataHub MCP Server `search`, `get_lineage`, `get_dataset_queries`, `get_entities`, `get_aspect_history`, `list_schema_fields`, `get_lineage_paths_between`, `add_tags`, `remove_tags`, `add_terms`, `remove_terms`, `add_owners`, `remove_owners`, `set_domains`, `remove_domains`, `update_description`, `add_structured_properties`, `remove_structured_properties`, `get_me`, `search_documents`, `grep_documents`, `save_document` ### Euno MCP `load_account_settings`, `list_personas`, `current_persona`, `switch_persona`, `load_skill`, `search_properties`, `fetch_single_resource`, `find_resource_by_name`, `find_resources_for_topic`, `get_upstream_lineage`, `resource_impact_analysis`, `query_to_eql`, `search_resources_eql`, `count_resources_eql`, `multi_count_resources_eql`, `pivot_resources_eql`, `search_resource_audit_log`, `sensitive_data_incident_response`, `documentation_search`, `documentation_get_full_document`, `documentation_get_surrounding_context`, `list_metadata_tags`, `list_metadata_tag_domains`, `preview_metadata_tag_plan`, `get_context_benchmark_overview`, `get_context_benchmark_questions`, `list_context_benchmark_runs`, `get_context_benchmark_run`, `get_context_benchmark_improvement_report`, `get_context_benchmark_question_result`, `get_context_benchmark_suggestion_application`, `get_context_benchmark_schedule`, `create_context_benchmark_question`, `update_context_benchmark_question`, `delete_context_benchmark_question`, `start_context_benchmark_run`, `stop_context_benchmark_run`, `update_context_benchmark_schedule`, `prepare_context_benchmark_suggestion_application`, `decide_context_benchmark_suggestion_application`, `undo_context_benchmark_suggestion`, `reapply_context_benchmark_suggestion`, `fetch_glossary_term`, `find_glossary_term`, `create_glossary_term`, `update_glossary_term`, `delete_glossary_term`, `fetch_metadata_tag`, `create_metadata_tag`, `update_metadata_tag`, `delete_metadata_tag`, `fetch_workflow`, `find_workflow`, `create_workflow`, `update_workflow`, `delete_workflow` **What Euno's server does not do.** Euno's server gives an agent governed context about the data; it does not run the query, which is the job of the warehouse's own server. ## Read next - [DataHub MCP Server, the full entry](/servers/datahub-mcp/) - [Euno MCP, the full entry](/servers/euno-mcp/) - [Every side-by-side](/compare/) ## Sources - raw.githubusercontent.com/acryldata/mcp-server-datahub/main/README.md: https://raw.githubusercontent.com/acryldata/mcp-server-datahub/main/README.md (read 10 Nov 2026) - pypi.org/pypi/mcp-server-datahub/json: https://pypi.org/pypi/mcp-server-datahub/json (read 10 Nov 2026) - github.com/acryldata/mcp-server-datahub/releases/tag/v0.8.0: https://github.com/acryldata/mcp-server-datahub/releases/tag/v0.8.0 (read 10 Nov 2026) - docs.euno.ai/ai/mcp-setup.md: https://docs.euno.ai/ai/mcp-setup.md (read 10 Nov 2026) - euno.ai: https://euno.ai/ (read 10 Nov 2026) - docs.euno.ai/support-resources/release-note/october-2026.md: https://docs.euno.ai/support-resources/release-note/october-2026.md (read 10 Nov 2026) - docs.euno.ai/ai/mcp-setup/mcp-bridge-setup.md: https://docs.euno.ai/ai/mcp-setup/mcp-bridge-setup.md (read 10 Nov 2026) - docs.euno.ai/ai/mcp-api-key-setup.md: https://docs.euno.ai/ai/mcp-api-key-setup.md (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/compare/datahub-mcp-vs-euno-mcp/ --- # Collibra MCP Server (chip) and Euno MCP, field by field: 49 and 56 tools named SIDE BY SIDE, read 10 Nov 2026 **The short answer.** Collibra MCP Server (chip) (Collibra, remote and local) names 49 tools; Euno MCP (Euno, remote (hosted)) names 56. Versions read on 10 Nov 2026: v0.0.67 and no version number published. The page puts the same fields in the same rows and draws no conclusion. | Field | Collibra MCP Server (chip) | Euno MCP | | --- | --- | --- | | Data tool | [Collibra](/mcp/collibra/) | [Euno](/mcp/euno/) | | Publisher | Collibra | Euno | | Kind | Official | Official | | Repository or docs | [github.com/collibra/chip](https://github.com/collibra/chip) | [docs.euno.ai/ai/mcp-setup](https://docs.euno.ai/ai/mcp-setup?utm_source=datamcpservers.com&utm_medium=referral&utm_campaign=side-by-side) | | Version or commit | V0.0.67 | Hosted service at api.app.euno.ai/mcp, no version published (docs read 10 Nov 2026) | | Last release | 27 Oct 2026 | none found on 10 Nov 2026 | | Transport | Remote and local | Remote (hosted) | | Auth | Collibra username/password: server-wide credentials over stdio (mcp.yaml or env vars), or client-supplied Basic Auth headers per request over HTTP; tools respect the connecting user's DGC scopes. | OAuth sign-in (default) or API key and headers; active persona scopes data and actions. | | Access | Configurable | Configurable | | Tools listed | 49 named in the evidence | 56 named in the evidence | | Tool definitions read at | [github.com/collibra/chip/blob/main/README.md](https://github.com/collibra/chip/blob/main/README.md) | [docs.euno.ai/ai/mcp-setup.md](https://docs.euno.ai/ai/mcp-setup.md?utm_source=datamcpservers.com&utm_medium=referral&utm_campaign=side-by-side) | | Context weight | 2,914 tokens for 41 tool definitions (41 with a description), measured from the README read | 760 tokens for 56 tool definitions (42 with a description), measured from the docs page read | | Status | Exists on 10 Nov 2026 as collibra/chip; v0.0.67 released 27 Oct 2026 per GitHub releases. Collibra's product-docs page for it returned a Cloudflare block, so only the repo README was read. | Exists on 10 Nov 2026; documented in docs.euno.ai MCP Setup Guide; tool names and descriptions published, parameter schemas not published. | | Read | 10 Nov 2026 | 10 Nov 2026 | The same fourteen fields for both servers, read 10 Nov 2026. ## In one sentence each - [Collibra MCP Server (chip)](/servers/collibra-mcp/): Collibra's server covers lineage with transformation logic, business terms and column semantics, classification, stewardship edits and data quality, under the connecting user's permissions. - [Euno MCP](/servers/euno-mcp/): Euno's hosted OAuth endpoint serves persona-scoped lineage, impact analysis, EQL search and catalog tools over its metadata graph, with no warehouse query tool. ## The read facts, topic by topic | Topic | Collibra MCP Server (chip) | Euno MCP | | --- | --- | --- | | Publisher and releases | The repository lives under the collibra GitHub organization and is titled the Collibra MCP Server (Apache-2.0). GitHub releases show v0.0.67 published 27 Oct 2026, so the server is still at a 0.0.x version. The release timestamp for v0.0.67 is 27 Oct 2026. | Euno publishes and documents its own MCP server at api.app.euno.ai/mcp. None found: A version number or dated release list for the Euno MCP server; the docs describe a hosted service, and the separate Euno SDK is v0.5. Stdio use goes through the third-party open-source mcp-remote bridge, not a Euno-published package. | | Clients, auth and scoping | The server is described as giving AI agents access to Collibra Data Governance Center for asset discovery, glossary queries and asset details. Permission scoping follows the connecting Collibra user: tools needing extra DGC scopes are flagged, and a missing scope returns a permission error. Natural-language discovery tools use Collibra AI and consume Collibra Units. With the HTTP transport, MCP clients are advised to supply their own Basic Auth headers per request; with stdio, one server-wide credential is used. | Euno MCP uses the active persona; if none can be identified the user is asked to choose. Euno MCP works with Cursor, Claude, ChatGPT in Developer Mode and Claude Code. Euno MCP has a dedicated Gemini Enterprise setup path besides direct, bridge and API-key setups. | | Access and safety | Data quality tools (18 tools, several that write or delete) are off by default and gated behind a --data-quality flag. Specific tools can be enabled or disabled through flags, environment variables or mcp.yaml; by default all are enabled. Destructive DQ tools use a confirm checkpoint (confirm=false by default is read-only). HTTP mode binds to localhost only, and credentials should be kept in environment variables. | Euno MCP includes read and write tools; write tools such as catalog creates and benchmark changes ask for client confirmation. Euno MCP catalog reads are published read-only and catalog writes are marked destructive; deletes need confirm=true. Euno enforces authentication and persona permissions on the server, independent of client confirmation. Context benchmark tools are accepted only for an account admin persona. | | What it reaches | None found: A list of warehouses, BI tools or transformation tools supported by the Collibra MCP README (coverage depends on what is catalogued in the Collibra instance). | The published tool table covers metadata, lineage, EQL, audit, documentation, catalog and benchmark tools; no listed tool executes SQL against a warehouse, and query_to_eql converts a filter without executing it. Euno's FAQ says Euno does not query customer data. | | Lineage and impact | The Collibra MCP server has separate tools for upstream and downstream technical lineage, lineage entities, and transformations. Get_lineage_transformation returns the details and logic of a specific data transformation. Lineage tools search entities and transformations in the technical lineage graph. Get_measure_data traces a measure back to its underlying physical columns and tables. None found: A README statement that the lineage tools return column-level edges or BI-field lineage explicitly; the tool list names entities and transformations only. | Euno MCP exposes get_upstream_lineage and resource_impact_analysis tools for one resource URI. Euno MCP can search and aggregate the metadata graph with EQL (search, count, multi-count, pivot). Euno MCP's example prompt traces a Tableau dashboard's upstream lineage. | | Definitions, ownership and rules | Get_business_term_data traces a business term back to its connected physical data assets. Get_column_semantics returns data attributes, measures and business assets connected to a column; get_table_semantics returns the table's semantic layer. Edit_asset can set attributes such as Definition and assign users or groups to resource roles such as Steward and Owner. Classification tools associate data classes with assets (add_data_classification_match, search_data_classification_match). | Euno MCP has glossary, metadata tag and workflow tools (14 direct catalog tools). Euno MCP's load_account_settings returns account, persona, skill and metadata context; skills carry organization-specific workflows. Euno MCP includes sensitive_data_incident_response returning evidence per user and BI-resource pair. | | Freshness and upkeep | None found: Any statement on collection cadence or freshness in the Collibra MCP README (the server queries a live Collibra instance). | The Euno MCP server keeps no session state, so connections survive restarts and scaling. Euno's October 2026 release notes add pivot and multi-count aggregation tools through MCP. | | Profile | The server is a Go binary distributed as prebuilt releases for Linux, macOS and Windows. | Euno's hosted MCP uses OAuth; an API-key setup exists for clients that cannot do interactive sign-in. | Each cell joins the facts read for that topic; the full list with sources is on each entry. Read 10 Nov 2026. ## Tool names, as published ### Collibra MCP Server (chip) `discover_business_glossary`, `discover_data_assets`, `get_assessment`, `get_asset_details`, `get_business_term_data`, `get_column_semantics`, `get_data_quality_rule`, `get_data_quality_rule_results`, `validate_data_quality_rule`, `list_data_quality_rule_templates`, `get_data_quality_rule_template`, `find_data_quality_rules`, `generate_data_quality_rule_sql`, `get_lineage_downstream`, `get_lineage_entity`, `get_lineage_transformation`, `get_lineage_upstream`, `get_measure_data`, `get_table_semantics`, `get_user_id_by_name`, `list_asset_types`, `list_data_contract`, `prepare_create_asset`, `pull_data_contract_manifest`, `search_asset_keyword`, `search_catalog_columns`, `search_data_class`, `search_data_classification_match`, `search_lineage_entities`, `search_lineage_transformations`, `add_data_classification_match`, `create_assessment`, `create_asset`, `create_data_quality_job`, `create_data_quality_rule`, `deploy_data_quality_rule_template`, `dq_cancel_job_run`, `dq_delete_job`, `dq_delete_job_run`, `dq_update_job`, `dq_get_job`, `dq_get_job_run`, `dq_search_jobs`, `dq_search_job_runs`, `edit_assessment`, `edit_asset`, `init_data_contract`, `push_data_contract_manifest`, `remove_data_classification_match` ### Euno MCP `load_account_settings`, `list_personas`, `current_persona`, `switch_persona`, `load_skill`, `search_properties`, `fetch_single_resource`, `find_resource_by_name`, `find_resources_for_topic`, `get_upstream_lineage`, `resource_impact_analysis`, `query_to_eql`, `search_resources_eql`, `count_resources_eql`, `multi_count_resources_eql`, `pivot_resources_eql`, `search_resource_audit_log`, `sensitive_data_incident_response`, `documentation_search`, `documentation_get_full_document`, `documentation_get_surrounding_context`, `list_metadata_tags`, `list_metadata_tag_domains`, `preview_metadata_tag_plan`, `get_context_benchmark_overview`, `get_context_benchmark_questions`, `list_context_benchmark_runs`, `get_context_benchmark_run`, `get_context_benchmark_improvement_report`, `get_context_benchmark_question_result`, `get_context_benchmark_suggestion_application`, `get_context_benchmark_schedule`, `create_context_benchmark_question`, `update_context_benchmark_question`, `delete_context_benchmark_question`, `start_context_benchmark_run`, `stop_context_benchmark_run`, `update_context_benchmark_schedule`, `prepare_context_benchmark_suggestion_application`, `decide_context_benchmark_suggestion_application`, `undo_context_benchmark_suggestion`, `reapply_context_benchmark_suggestion`, `fetch_glossary_term`, `find_glossary_term`, `create_glossary_term`, `update_glossary_term`, `delete_glossary_term`, `fetch_metadata_tag`, `create_metadata_tag`, `update_metadata_tag`, `delete_metadata_tag`, `fetch_workflow`, `find_workflow`, `create_workflow`, `update_workflow`, `delete_workflow` **What Euno's server does not do.** Euno's server gives an agent governed context about the data; it does not run the query, which is the job of the warehouse's own server. ## Read next - [Collibra MCP Server (chip), the full entry](/servers/collibra-mcp/) - [Euno MCP, the full entry](/servers/euno-mcp/) - [Every side-by-side](/compare/) ## Sources - github.com/collibra/chip/blob/main/README.md: https://github.com/collibra/chip/blob/main/README.md (read 10 Nov 2026) - github.com/collibra/chip/releases: https://github.com/collibra/chip/releases (read 10 Nov 2026) - docs.euno.ai/ai/mcp-setup.md: https://docs.euno.ai/ai/mcp-setup.md (read 10 Nov 2026) - euno.ai: https://euno.ai/ (read 10 Nov 2026) - docs.euno.ai/support-resources/release-note/october-2026.md: https://docs.euno.ai/support-resources/release-note/october-2026.md (read 10 Nov 2026) - docs.euno.ai/ai/mcp-setup/mcp-bridge-setup.md: https://docs.euno.ai/ai/mcp-setup/mcp-bridge-setup.md (read 10 Nov 2026) - docs.euno.ai/ai/mcp-api-key-setup.md: https://docs.euno.ai/ai/mcp-api-key-setup.md (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/compare/collibra-mcp-vs-euno-mcp/ --- # 20 pairs of MCP servers, the same fields in the same rows SIDE BY SIDE, read 10 Nov 2026 **The short answer.** The directory publishes 20 side-by-side pages: the three Snowflake servers against each other, the two for BigQuery and for Databricks, and pairs engineers read together, such as Atlan and DataHub or Monte Carlo and SYNQ. Each puts the read facts in the same rows and draws no conclusion. | Side by side | Data tools | Tools named | | --- | --- | --- | | [Airbyte MCP (hosted) and Fivetran MCP Server](/compare/airbyte-mcp-vs-fivetran-mcp/) | Airbyte, Fivetran | 50 and 53 | | [Alation AI Agent SDK MCP server and Collibra MCP Server (chip)](/compare/alation-mcp-vs-collibra-mcp/) | Alation, Collibra | 12 and 49 | | [Atlan MCP Server and DataHub MCP Server](/compare/atlan-mcp-vs-datahub-mcp/) | Atlan, DataHub | 41 and 22 | | [Atlan MCP Server and Euno MCP](/compare/atlan-mcp-vs-euno-mcp/) | Atlan, Euno | 41 and 56 | | [BigQuery remote MCP server and MCP Toolbox for Databases (BigQuery tools)](/compare/bigquery-remote-mcp-vs-mcp-toolbox-bigquery/) | BigQuery, BigQuery | 9 and 9 | | [Collibra MCP Server (chip) and Euno MCP](/compare/collibra-mcp-vs-euno-mcp/) | Collibra, Euno | 49 and 56 | | [Cube MCP server and dbt MCP Server](/compare/cube-mcp-vs-dbt-mcp/) | Cube, dbt | 30 and 52 | | [Databricks Labs MCP servers and Databricks managed MCP servers](/compare/databrickslabs-mcp-vs-databricks-managed-mcp/) | Databricks, Databricks | 3 and 4 | | [DataHub MCP Server and Euno MCP](/compare/datahub-mcp-vs-euno-mcp/) | DataHub, Euno | 22 and 56 | | [DataHub MCP Server and OpenMetadata MCP Server](/compare/datahub-mcp-vs-openmetadata-mcp/) | DataHub, OpenMetadata | 22 and 15 | | [Hex MCP server and Omni MCP Server](/compare/hex-mcp-vs-omni-mcp/) | Hex, Omni | 18 and 26 | | [Looker tools in MCP Toolbox for Databases and Tableau MCP](/compare/looker-mcp-toolbox-vs-tableau-mcp/) | Looker, Tableau | 55 and 40 | | [mcp-snowflake-server (community) and Snowflake Labs MCP Server](/compare/isaacwasserman-mcp-snowflake-server-vs-snowflake-labs-mcp/) | Snowflake, Snowflake | 8 and 6 | | [mcp-snowflake-server (community) and Snowflake-managed MCP server](/compare/isaacwasserman-mcp-snowflake-server-vs-snowflake-managed-mcp/) | Snowflake, Snowflake | 8 and 6 | | [Monte Carlo MCP Server and Sifflet MCP Server](/compare/monte-carlo-mcp-vs-sifflet-mcp/) | Monte Carlo, Sifflet | 15 and 10 | | [Monte Carlo MCP Server and SYNQ MCP (Coalesce Quality)](/compare/monte-carlo-mcp-vs-synq-mcp/) | Monte Carlo, SYNQ | 15 and 134 | | [Power BI Modeling MCP Server and Tableau MCP](/compare/powerbi-modeling-mcp-vs-tableau-mcp/) | Power BI, Tableau | 21 and 40 | | [Secoda MCP and Select Star MCP Server](/compare/secoda-mcp-vs-select-star-mcp/) | Secoda, Select Star | 8 and 7 | | [Sigma MCP server and ThoughtSpot MCP Server](/compare/sigma-mcp-vs-thoughtspot-mcp-server/) | Sigma, ThoughtSpot | 0 and 20 | | [Snowflake Labs MCP Server and Snowflake-managed MCP server](/compare/snowflake-labs-mcp-vs-snowflake-managed-mcp/) | Snowflake, Snowflake | 6 and 6 | Filed alphabetically by the first server's name, read 10 Nov 2026. Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/compare/ --- # When analysts stop trusting the dashboards, check which servers can trace a number and report incidents SITUATION, read 10 Nov 2026 **The short answer.** A data platform lead whose analysts distrust the dashboards needs an agent that can trace a number upstream and see open incidents. Nineteen catalog, observability and BI servers are in scope here; the checks below name the fields that answer each question, from lineage tools to auth, read 10 Nov 2026. The question behind an untrusted dashboard is where the number came from and whether something upstream broke. An agent can only help if a server gives it lineage or impact tools, and, for incidents, an observability server. The page lists what to check and the published facts for each server in scope. It names no choice: the facts differ by stack, and the engineer's own stack decides which server is relevant. ## What to check | Check | The field or fact that answers it | | --- | --- | | Can the agent follow a BI field back to its source tables? | Lineage and impact facts on each entry; tool names such as get_lineage or get_upstream_lineage | | Can the agent see open data incidents? | Observability servers' tool names and the Access field | | Does the agent see only what the person asking may see? | Auth field: OAuth as the user versus a shared token | | Is the server current? | Version or commit and Last release fields | ## The published facts, for the 19 servers in scope | Server | Kind | Transport | Auth | Tools listed | Last release | | --- | --- | --- | --- | --- | --- | | [Alation AI Agent SDK MCP server](/servers/alation-mcp/) | Official | Remote and local | OAuth client credentials of a service account (role-limited) or user refresh token; remote MCP at /ai/mcp needs client_id and client_secret (no dynamic client registration). | 12 named in the evidence | 26 Oct 2026 | | [Atlan MCP Server](/servers/atlan-mcp/) | Official | Remote (hosted) | OAuth per user (calls run as the signed-in user) or Atlan API key as Bearer token (shared service identity); local server uses API key + base URL. | 41 named in the evidence | 17 Feb 2026 | | [Collibra MCP Server (chip)](/servers/collibra-mcp/) | Official | Remote and local | Collibra username/password: server-wide credentials over stdio (mcp.yaml or env vars), or client-supplied Basic Auth headers per request over HTTP; tools respect the connecting user's DGC scopes. | 49 named in the evidence | 27 Oct 2026 | | [Cube MCP server](/servers/cube-mcp/) | Official | Remote (hosted) | OAuth Authorization Code + PKCE (scope mcp-agent-access, public client cube-mcp-client), or a Cube API key exchanged for a token (client_id cube-api-key). | 30 named in the evidence | none found on 10 Nov 2026 | | [DataHub MCP Server](/servers/datahub-mcp/) | Official | Remote and local | Stdio: DATAHUB_GMS_URL + DATAHUB_GMS_TOKEN (or ~/.datahubenv); HTTP: per-request DataHub personal access token as Authorization Bearer, no shared server token. | 22 named in the evidence | 21 Oct 2026 | | [Euno MCP](/servers/euno-mcp/) | Official | Remote (hosted) | OAuth sign-in (default) or API key and headers; active persona scopes data and actions. | 56 named in the evidence | none found on 10 Nov 2026 | | [Hex MCP server](/servers/hex-mcp/) | Official | Remote (hosted) | OAuth authorization to the user's Hex workspace (server URL app.hex.tech/mcp, or the customer's single-tenant, EU or HIPAA domain); requires Explorer role or higher. | 18 named in the evidence | none found on 10 Nov 2026 | | [Looker tools in MCP Toolbox for Databases](/servers/looker-mcp-toolbox/) | Official | Remote and local | Looker API client_id and client_secret in the Looker source config, or use_client_oauth to pass the caller's OAuth token (default false). | 55 named in the evidence | 5 Nov 2026 | | [Monte Carlo MCP Server](/servers/monte-carlo-mcp/) | Official | Remote (hosted) | OAuth 2.1 with Dynamic Client Registration (user identity, Editor role or above), or MCP-server-type API key id:secret as Bearer token via mcp-remote. | 15 named in the evidence | none found on 10 Nov 2026 | | [Omni MCP Server](/servers/omni-mcp/) | Official | Remote (hosted) | OAuth 2.1 with PKCE (creates an MCP OAuth PAT for the user), or an API key (personal token or Organization API key; Organization keys can act as a user with X-MCP-User-ID). | 26 named in the evidence | none found on 10 Nov 2026 | | [OpenMetadata MCP Server](/servers/openmetadata-mcp/) | Official | Remote (hosted) | OAuth 2.0 via existing SSO (per user), personal access token, or Bot JWT as Authorization Bearer. | 15 named in the evidence | 27 Oct 2026 | | [Power BI Modeling MCP Server](/servers/powerbi-modeling-mcp/) | Official | Remote and local | Local: Microsoft Entra interactive sign-in (default) or service principal (client secret or certificate). Hosted: Entra ID as the signed-in user. | 21 named in the evidence | 6 Nov 2026 | | [Secoda MCP](/servers/secoda-mcp/) | Official | Remote and local | Secoda API token (API_TOKEN env var) for the local FastMCP wrapper; docs also describe a Secoda-hosted MCP endpoint authenticated against workspace and AI permissions. | 8 named in the evidence | none found on 10 Nov 2026 | | [Select Star MCP Server](/servers/select-star-mcp/) | Official | Remote (hosted) | Select Star API token sent as Authorization: Bearer header (mcp-remote proxy used for Claude Desktop). | 7 named in the evidence | none found on 10 Nov 2026 | | [Sifflet MCP Server](/servers/sifflet-mcp/) | Official | Local (stdio) | Sifflet API token (Viewer role for most tools, Editor for monitor generation) and backend URL as environment variables. | 10 named in the evidence | 3 Aug 2026 | | [Sigma MCP server](/servers/sigma-mcp/) | Official | Remote (hosted) | OAuth (HTTP transport) inheriting the user's Sigma account permissions; API-key access token with scope mcp:access (Beta). | 0 named in the evidence | none found on 10 Nov 2026 | | [SYNQ MCP (Coalesce Quality)](/servers/synq-mcp/) | Official | Remote (hosted) | OAuth 2 with PKCE per user with selectable permission tier (read, write, governance, deploy); legacy workspace token (st-...) as Bearer. | 134 named in the evidence | none found on 10 Nov 2026 | | [Tableau MCP](/servers/tableau-mcp/) | Official | Remote and local | Hosted: OAuth 2.1 with each user's own Tableau Cloud identity. Self-hosted: PAT (default), direct-trust (Connected Apps), OAuth, passthrough, or unified access tokens. | 40 named in the evidence | 9 Nov 2026 | | [ThoughtSpot MCP Server](/servers/thoughtspot-mcp-server/) | Official | Remote (hosted) | OAuth to a ThoughtSpot instance (dynamic client registration, or manually registered client id and secret); bearer-token and /sse paths also exist; the user chooses or is bound to a ThoughtSpot instance URL. | 20 named in the evidence | 6 Jun 2025 | Filed by data tool, read 10 Nov 2026. Scope: catalogs and metadata, data observability, bi and analytics. ## What each server documents on lineage and impact - [Alation AI Agent SDK MCP server](/servers/alation-mcp/): Alation's lineage tool traverses upstream or downstream from a starting object and supports column-level lineage; it is a BETA feature that Alation support must enable on the instance. The lineage tool takes max_depth (default 10), allowed_otypes, design/run-time filters and schema exclusions. None found: Documentation that the lineage tool returns transformation SQL/logic or BI-field/metric nodes; the tool page lists graph nodes filtered by object type only. - [Atlan MCP Server](/servers/atlan-mcp/): The hosted Atlan MCP exposes a traverse_lineage tool for upstream/downstream lineage with deeper hops on request. The docs say the hosted Atlan MCP can traverse end-to-end lineage; the tool page does not state column-level or SQL-logic parsing for the lineage tool. None found: A statement on the MCP tools page that traverse_lineage returns column-level edges or transformation SQL. Query_assets runs read-only SQL against a connected warehouse (Snowflake, Redshift, BigQuery, Databricks), returning up to 100 rows. - [Collibra MCP Server (chip)](/servers/collibra-mcp/): The Collibra MCP server has separate tools for upstream and downstream technical lineage, lineage entities, and transformations. Get_lineage_transformation returns the details and logic of a specific data transformation. Lineage tools search entities and transformations in the technical lineage graph. Get_measure_data traces a measure back to its underlying physical columns and tables. None found: A README statement that the lineage tools return column-level edges or BI-field lineage explicitly; the tool list names entities and transformations only. - [Cube MCP server](/servers/cube-mcp/): SearchDataModel searches the semantic model by similarity, covering views and their measures and dimensions, and returns name, title, description and type. ListDataModelFiles and readDataModelFile read the raw YAML, JavaScript or Python source of the semantic model, so the definition logic itself is readable. None found: A lineage tool in the Cube MCP tool tables (getBranchDiff and getDataModelChanges compare model versions, not data lineage). - [DataHub MCP Server](/servers/datahub-mcp/): DataHub's MCP server exposes a get_lineage tool covering datasets, columns and dashboards with hop control. The README says lineage is traceable at table and column level and that transformation logic is understood. A get_lineage_paths_between tool returns exact lineage paths between two assets or columns, including intermediate transformations and SQL query information. Get_dataset_queries fetches real SQL queries that reference a dataset or column. - [Euno MCP](/servers/euno-mcp/): Euno MCP exposes get_upstream_lineage and resource_impact_analysis tools for one resource URI. Euno MCP can search and aggregate the metadata graph with EQL (search, count, multi-count, pivot). Euno MCP's example prompt traces a Tableau dashboard's upstream lineage. - [Hex MCP server](/servers/hex-mcp/): Project editing tools list cells and read cell outputs and chart images, giving agents the SQL, Python and Markdown cells of a notebook. Search_projects finds projects in the workspace, and get_project reads a project's metadata. None found: A lineage tool on the Hex MCP server page (cell-level reads exist, but no column or upstream-table lineage tool is listed). - [Looker tools in MCP Toolbox for Databases](/servers/looker-mcp-toolbox/): Looker-get-explore returns explore metadata including always_filter and conditionally_filter constraints, tags and hidden status. Separate tools return dimensions, measures, filters and parameters of an explore, and the columns of connection tables. Looker-get-project-file returns the contents of a LookML file, so an agent can read the transformation and field logic written in LookML. None found: A lineage tool among the 55 looker-* tools in the Toolbox llms.txt index (no column-lineage or upstream-table tool is named). - [Monte Carlo MCP Server](/servers/monte-carlo-mcp/): Monte Carlo's MCP server includes a get_asset_lineage tool, used to see how data flows into a table. None found: Documentation that get_asset_lineage returns column-level lineage, transformation SQL, BI fields or metrics; the MCP pages name only asset lineage. - [Omni MCP Server](/servers/omni-mcp/): PickModel returns the models available to the calling user and pickTopic returns the topics in the chosen model and selects the most relevant one. SearchContent finds existing dashboards and apps by name, folder or tile name, and listDocumentQueries lists the saved queries in a document. None found: A lineage tool among the Omni MCP tools (no column or upstream-table lineage tool documented). - [OpenMetadata MCP Server](/servers/openmetadata-mcp/): Get_entity_lineage returns upstream and downstream lineage for any entity, with an includeColumnLineage option (default false, table level only), up to 10 hops. Create_lineage lets an agent write lineage edges, and root_cause_analysis is a lineage-and-impact tool. None found: Documentation that the MCP lineage tools return transformation SQL or logic. - [Power BI Modeling MCP Server](/servers/powerbi-modeling-mcp/): The server exposes model-object tools for tables, columns, measures, relationships, partitions, hierarchies, calculation groups and named expressions (model definition level). Database_operations can import and export TMDL folders and deploy to Fabric, giving agents the model definition as code. None found: A lineage tool in the Power BI Authoring MCP tool table (no upstream-table, column-lineage or report-dependency tool). - [Secoda MCP](/servers/secoda-mcp/): The entity_lineage tool explores upstream and downstream dependencies for an entity. In the open-source wrapper entity_lineage takes only an entity_id and no depth or column parameter. None found: Documentation of column-level lineage, transformation SQL, BI-field or metric lineage in the Secoda MCP docs. The run_sql tool executes SQL against connected warehouses. - [Select Star MCP Server](/servers/select-star-mcp/): The Select Star MCP server has a traverse_lineage tool for lineage relationships and dependencies. Get_asset_queries returns warehouse queries connected to tables or columns. None found: Documentation on the MCP page that traverse_lineage is column-level or returns transformation SQL, BI-field or metric lineage. - [Sifflet MCP Server](/servers/sifflet-mcp/): Sifflet MCP exposes get_downstream_assets_of_asset, which returns all downstream assets of an asset by URN. Sifflet says the server can start from an incident and trace the downstream assets affected. - [Sigma MCP server](/servers/sigma-mcp/): Agents can explore data sources and documents by asking about column names, definitions, descriptions, data types and column attributes. Search covers documents and data sources including data models, warehouse tables, and elements from workbooks and data models. None found: Lineage or upstream-dependency capability on the Sigma MCP server page. - [SYNQ MCP (Coalesce Quality)](/servers/synq-mcp/): SYNQ (docs now titled Coalesce Quality) serves the same lineage graph as its lineage view, including column-level lineage where the warehouse supports it. Lineage tools include get_lineage (up/down/both with depth), upstream and downstream dependencies, common upstream, and upstream sources such as dbt sources. Get_code returns the latest source code defining an entity, including SQL definitions and dbt models. Impact tools list the entities and systems affected by an entity or an issue. - [Tableau MCP](/servers/tableau-mcp/): Tableau's get-datasource-metadata tool returns table relationships, field descriptions, roles and types, calculation strings and parameters for a data source (table, column and calculation-definition level, not transformation lineage). The describe-flow tool summarizes a Tableau Prep flow's design including inputs, outputs, steps, lineage and connections, on all SKUs. On Tableau+ only, query-knowledge-context queries governed definitions, relationships, lineage, impact and graph sources through one read-only tool. The server's get-datasource-metadata tool enriches VizQL Data Service results with Tableau's Metadata API and only returns the richer descriptions for published data sources. - [ThoughtSpot MCP Server](/servers/thoughtspot-mcp-server/): Search_objects finds Answers, Liveboards and Worksheets with filters for types, owner, tag, modified_since and verified_only, returning object metadata (title, description, query tokens, tags). The tool source warns that description and query tokens are authoring metadata that can be stale, and advises grounding explanations in the object's data via get_data. None found: A lineage tool in tool-definitions.ts (no column, upstream-table or transformation lineage tool among the 20 tools). **What Euno's server does not do.** Euno's server gives an agent governed context about the data; it does not run the query, which is the job of the warehouse's own server. ## Other situations - [An analytics engineering team on dbt and Snowflake or Databricks](/situations/dbt-warehouse/) - [A company putting AI agents on its data this year](/situations/agents-this-year/) - [A governance lead who must show where a regulated field flows](/situations/regulated-field/) - [A BI team with thousands of reports and no map of what feeds them](/situations/bi-sprawl/) ## Sources - github.com/Alation/alation-ai-agent-sdk/blob/main/guides/tools/lineage.md: https://github.com/Alation/alation-ai-agent-sdk/blob/main/guides/tools/lineage.md (read 10 Nov 2026) - docs.atlan.com/product/capabilities/atlan-ai/references/mcp-tools: https://docs.atlan.com/product/capabilities/atlan-ai/references/mcp-tools (read 10 Nov 2026) - docs.atlan.com/product/capabilities/atlan-ai/how-tos/remote-mcp-overview: https://docs.atlan.com/product/capabilities/atlan-ai/how-tos/remote-mcp-overview (read 10 Nov 2026) - github.com/collibra/chip/blob/main/README.md: https://github.com/collibra/chip/blob/main/README.md (read 10 Nov 2026) - docs.cube.dev/docs/integrations/mcp-server.md: https://docs.cube.dev/docs/integrations/mcp-server.md (read 10 Nov 2026) - raw.githubusercontent.com/acryldata/mcp-server-datahub/main/README.md: https://raw.githubusercontent.com/acryldata/mcp-server-datahub/main/README.md (read 10 Nov 2026) - docs.euno.ai/ai/mcp-setup.md: https://docs.euno.ai/ai/mcp-setup.md (read 10 Nov 2026) - learn.hex.tech/docs/api-integrations/mcp-server: https://learn.hex.tech/docs/api-integrations/mcp-server (read 10 Nov 2026) - mcp-toolbox.dev/integrations/looker/tools/looker-get-explore: https://mcp-toolbox.dev/integrations/looker/tools/looker-get-explore/ (read 10 Nov 2026) - mcp-toolbox.dev/llms.txt: https://mcp-toolbox.dev/llms.txt (read 10 Nov 2026) - docs.getmontecarlo.com/docs/mcp-server: https://docs.getmontecarlo.com/docs/mcp-server (read 10 Nov 2026) - docs.omni.co/ai/mcp/tools.md: https://docs.omni.co/ai/mcp/tools.md (read 10 Nov 2026) - docs.open-metadata.org/v2.0.x/how-to-guides/mcp/reference.md: https://docs.open-metadata.org/v2.0.x/how-to-guides/mcp/reference.md (read 10 Nov 2026) - raw.githubusercontent.com/microsoft/powerbi-modeling-mcp/main/README.md: https://raw.githubusercontent.com/microsoft/powerbi-modeling-mcp/main/README.md (read 10 Nov 2026) - docs.secoda.co/features/ai-assistant/secoda-mcp-server.md: https://docs.secoda.co/features/ai-assistant/secoda-mcp-server.md (read 10 Nov 2026) - github.com/secoda/secoda-mcp/blob/main/server.py: https://github.com/secoda/secoda-mcp/blob/main/server.py (read 10 Nov 2026) - docs.selectstar.com/features/mcp-server: https://docs.selectstar.com/features/mcp-server (read 10 Nov 2026) - github.com/siffletdata/sifflet-mcp/blob/1f390ca06d7919214bfe067991539502fdb4b9fc/src/sifflet_mcp/server.py: https://github.com/siffletdata/sifflet-mcp/blob/1f390ca06d7919214bfe067991539502fdb4b9fc/src/sifflet_mcp/server.py (read 10 Nov 2026) - github.com/siffletdata/sifflet-mcp/blob/main/README.md: https://github.com/siffletdata/sifflet-mcp/blob/main/README.md (read 10 Nov 2026) - help.sigmacomputing.com/docs/use-sigma-mcp-server: https://help.sigmacomputing.com/docs/use-sigma-mcp-server (read 10 Nov 2026) - docs.synq.io/scout/mcp.md: https://docs.synq.io/scout/mcp.md (read 10 Nov 2026) - tableau.github.io/tableau-mcp/docs/intro: https://tableau.github.io/tableau-mcp/docs/intro (read 10 Nov 2026) - raw.githubusercontent.com/tableau/tableau-mcp/main/src/tools/web/getDatasourceMetadata/getDatasourceMetadata.ts: https://raw.githubusercontent.com/tableau/tableau-mcp/main/src/tools/web/getDatasourceMetadata/getDatasourceMetadata.ts (read 10 Nov 2026) - raw.githubusercontent.com/thoughtspot/mcp-server/main/src/servers/tool-definitions.ts: https://raw.githubusercontent.com/thoughtspot/mcp-server/main/src/servers/tool-definitions.ts (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/situations/untrusted-dashboards/ --- # On dbt with Snowflake or Databricks, six servers cover the stack, and they split SQL, models and semantics differently SITUATION, read 10 Nov 2026 **The short answer.** An analytics engineering team on dbt with Snowflake or Databricks has six servers to read: the dbt MCP Server, three for Snowflake and two for Databricks. The checks below cover which ones run SQL, which read models and the semantic layer, how each signs in, and which are deprecated, read 10 Nov 2026. A dbt team wires an agent to two places: the dbt project (models, tests, lineage, metrics) and the warehouse that runs it. The dbt MCP Server covers the first; the warehouse servers cover the second. Two of the six carry a deprecation notice in their own README: the Snowflake Labs server and the Databricks Labs Unity Catalog server. Both point to the vendor's managed server. ## What to check | Check | The field or fact that answers it | | --- | --- | | Which server runs SQL, and as whom? | Auth and Access fields; SQL tool names | | Which server reads dbt models, tests and column lineage? | The dbt MCP Server's tool names and lineage facts | | Is the server deprecated or moved? | Status field | | Hosted or local? | Transport field | ## The published facts, for the 9 servers in scope | Server | Kind | Transport | Auth | Access | Status | | --- | --- | --- | --- | --- | --- | | [Amazon Redshift MCP Server (AWS Labs)](/servers/awslabs-redshift-mcp-server/) | Official, Labs org | Local (stdio) | AWS credentials (profile/env) with IAM and Redshift Data API; database user privileges. | Configurable | Exists in awslabs/mcp; PyPI 0.1.3 uploaded 9 Nov 2026 (read 10 Nov 2026). | | [BigQuery remote MCP server](/servers/bigquery-remote-mcp/) | Official | Remote (hosted) | OAuth 2.0 with IAM (Google Cloud identities, OAuth client, or agent identity). | Configurable | Exists; Google-managed endpoint, reference page last updated 3 Nov 2026 (read 10 Nov 2026). | | [MCP Toolbox for Databases (BigQuery tools)](/servers/mcp-toolbox-bigquery/) | Official | Remote and local | Application Default Credentials, client OAuth token forwarding, or service account impersonation. | Configurable | Repo googleapis/genai-toolbox now lives at googleapis/mcp-toolbox (read 10 Nov 2026); BigQuery tools documented there. | | [Databricks Labs MCP servers](/servers/databrickslabs-mcp/) | Official, Labs org | Remote and local | Databricks CLI-style credentials; Unity Catalog permissions. | Not stated in the source read | Exists; README marks Unity Catalog server deprecated in favor of managed MCP servers; last push 26 Oct 2026 (GitHub API); no tagged release read. | | [Databricks managed MCP servers](/servers/databricks-managed-mcp/) | Official | Remote (hosted) | Databricks workspace sign-in; caller's Unity Catalog permissions; Unity Gateway grants and policies. | Configurable | Documented by Databricks as Databricks-provided MCPs (read 10 Nov 2026); no versioned release stream found. | | [dbt MCP Server](/servers/dbt-mcp/) | Official | Remote and local | dbt service/personal token and environment IDs for platform tools; OAuth or token per docs; local dbt project for CLI tools. | Configurable | Exists, v2.10.1 on 3 Nov 2026 (PyPI and GitHub). | | [mcp-snowflake-server (community)](/servers/isaacwasserman-mcp-snowflake-server/) | Community | Local (stdio) | Snowflake user credentials (password, private key, external browser) or connections.toml. | Configurable | Exists as a community server; PyPI 0.4.0 dated 10 Apr 2025, GitHub pushed 7 Oct 2025 (GitHub API). | | [Snowflake Labs MCP Server](/servers/snowflake-labs-mcp/) | Official, Labs org | Remote and local | Snowflake Python connector: password/PAT, key pair, OAuth, SSO, MFA; honors the connection role's RBAC. | Configurable | Exists, but the README states it is deprecated and no longer maintained and points to the Snowflake-managed MCP server (read 10 Nov 2026). | | [Snowflake-managed MCP server](/servers/snowflake-managed-mcp/) | Official | Remote (hosted) | Snowflake OAuth (default) or External OAuth; PAT supported; RBAC per MCP server and per tool. | Configurable | Documented by Snowflake as the Snowflake-managed MCP server (read 10 Nov 2026); the Snowflake-Labs repository points users here. | Filed by data tool, read 10 Nov 2026. Scope: warehouses, transformation. ## What each server documents on publisher and releases - [Amazon Redshift MCP Server (AWS Labs)](/servers/awslabs-redshift-mcp-server/): The server lives in the awslabs/mcp repository and PyPI lists author Amazon Web Services; awslabs.redshift-mcp-server 0.1.3 was uploaded 9 Nov 2026. README calls it a server to discover, explore and query Redshift clusters and serverless workgroups. - [BigQuery remote MCP server](/servers/bigquery-remote-mcp/): Google documents a BigQuery remote MCP server, enabled when the BigQuery API is enabled. The reference page for bigquery.googleapis.com MCP shows Last updated 3 Nov 2026 UTC; no semantic version is published. - [MCP Toolbox for Databases (BigQuery tools)](/servers/mcp-toolbox-bigquery/): MCP Toolbox for Databases is an open source MCP server in the googleapis organization (formerly genai-toolbox) with a BigQuery prebuilt config. The latest GitHub release is v1.16.0, whose release timestamp on the page is 5 Nov 2026. - [Databricks Labs MCP servers](/servers/databrickslabs-mcp/): Databrickslabs/mcp is a Databricks Labs repository whose README calls the Unity Catalog server deprecated in favor of Managed MCP servers. The repo has no GitHub release entries read; GitHub API shows last push 26 Oct 2026 and no archive flag. - [Databricks managed MCP servers](/servers/databricks-managed-mcp/): Databricks documents Databricks-provided MCPs in the system.ai schema, with no server to host or register; the page was last updated Nov 9, 2026. Genie One MCP is generally available while dbsql, sandbox and web_search require the Unity Gateway beta. - [dbt MCP Server](/servers/dbt-mcp/): dbt Labs publishes dbt-mcp; the latest GitHub release is v2.10.1, whose release timestamp on the page is 3 Nov 2026 (PyPI also 3 Nov 2026). dbt Labs documents the MCP server on docs.getdbt.com and ships it as self-hosted or remote. - [mcp-snowflake-server (community)](/servers/isaacwasserman-mcp-snowflake-server/): The server is a community project by isaacwasserman, not published by Snowflake; license is GPL-3.0 per GitHub API. PyPI lists mcp-snowflake-server 0.4.0 uploaded 10 Apr 2025; the GitHub repo was last pushed 7 Oct 2025. - [Snowflake Labs MCP Server](/servers/snowflake-labs-mcp/): Snowflake Labs marks its Snowflake-Labs/mcp repository as deprecated and unmaintained, pointing users to the official Snowflake MCP server. The latest PyPI/GitHub release of snowflake-labs-mcp is v1.4.2, published 15 May 2026 and titled as the release that starts deprecation. PyPI lists snowflake-labs-mcp 1.4.2 uploaded 15 May 2026 by Snowflake Labs authors. - [Snowflake-managed MCP server](/servers/snowflake-managed-mcp/): Snowflake documents a Snowflake-managed MCP server that lets agents retrieve data from Snowflake accounts without deploying separate infrastructure. The docs state support for MCP protocol revision 25 Nov 2025; no semantic version number or release date is published on the page. ## Other situations - [A data platform lead whose analysts no longer trust the dashboards](/situations/untrusted-dashboards/) - [A company putting AI agents on its data this year](/situations/agents-this-year/) - [A governance lead who must show where a regulated field flows](/situations/regulated-field/) - [A BI team with thousands of reports and no map of what feeds them](/situations/bi-sprawl/) ## Sources - pypi.org/pypi/awslabs.redshift-mcp-server/json: https://pypi.org/pypi/awslabs.redshift-mcp-server/json (read 10 Nov 2026) - raw.githubusercontent.com/awslabs/mcp/main/src/redshift-mcp-server/README.md: https://raw.githubusercontent.com/awslabs/mcp/main/src/redshift-mcp-server/README.md (read 10 Nov 2026) - docs.cloud.google.com/bigquery/docs/use-bigquery-mcp: https://docs.cloud.google.com/bigquery/docs/use-bigquery-mcp (read 10 Nov 2026) - docs.cloud.google.com/bigquery/docs/reference/mcp: https://docs.cloud.google.com/bigquery/docs/reference/mcp (read 10 Nov 2026) - raw.githubusercontent.com/googleapis/mcp-toolbox/main/README.md: https://raw.githubusercontent.com/googleapis/mcp-toolbox/main/README.md (read 10 Nov 2026) - github.com/googleapis/mcp-toolbox/releases: https://github.com/googleapis/mcp-toolbox/releases (read 10 Nov 2026) - raw.githubusercontent.com/databrickslabs/mcp/master/README.md: https://raw.githubusercontent.com/databrickslabs/mcp/master/README.md (read 10 Nov 2026) - docs.databricks.com/aws/en/generative-ai/mcp/managed-mcp: https://docs.databricks.com/aws/en/generative-ai/mcp/managed-mcp (read 10 Nov 2026) - github.com/dbt-labs/dbt-mcp/releases: https://github.com/dbt-labs/dbt-mcp/releases (read 10 Nov 2026) - docs.getdbt.com/docs/dbt-ai/about-mcp: https://docs.getdbt.com/docs/dbt-ai/about-mcp (read 10 Nov 2026) - raw.githubusercontent.com/isaacwasserman/mcp-snowflake-server/main/README.md: https://raw.githubusercontent.com/isaacwasserman/mcp-snowflake-server/main/README.md (read 10 Nov 2026) - pypi.org/pypi/mcp-snowflake-server/json: https://pypi.org/pypi/mcp-snowflake-server/json (read 10 Nov 2026) - raw.githubusercontent.com/Snowflake-Labs/mcp/main/README.md: https://raw.githubusercontent.com/Snowflake-Labs/mcp/main/README.md (read 10 Nov 2026) - github.com/Snowflake-Labs/mcp/releases: https://github.com/Snowflake-Labs/mcp/releases (read 10 Nov 2026) - pypi.org/pypi/snowflake-labs-mcp/json: https://pypi.org/pypi/snowflake-labs-mcp/json (read 10 Nov 2026) - docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-mcp: https://docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-mcp (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/situations/dbt-warehouse/ --- # Putting agents on the data this year: check transport, auth, write access and tool-definition size, server by server SITUATION, read 10 Nov 2026 **The short answer.** A company putting agents on its data this year will connect several servers at once, so transport, auth as the user, write access and the size of each server's tool definitions all matter. All 30 servers are in scope; the table shows those fields side by side, read 10 Nov 2026. An agent that reaches the warehouse, the transformation layer, the BI tool and the catalog loads every server's tool definitions into its context window. The Context Weight Index measures that size from each server's published definitions. Hosted servers with OAuth sign each person in as themselves; local servers usually run with a token on the user's machine. The table shows which is which. ## What to check | Check | The field or fact that answers it | | --- | --- | | Hosted endpoint or a process on each laptop? | Transport field | | Does each call run as the person asking? | Auth field | | Can the agent change anything? | Access field and write tool names | | How much context do the tool definitions take? | Context weight field and the Context Weight Index | ## The published facts, for the 30 servers in scope | Server | Transport | Auth | Access | Context weight | | --- | --- | --- | --- | --- | | [Airbyte MCP (hosted)](/servers/airbyte-mcp/) | Remote (hosted) | OAuth (Streamable HTTP) as the signed-in Airbyte user, or an Airbyte application for non-browser clients; tools run with that user's Airbyte role. | Configurable | 870 tokens for 50 tool definitions (50 with a description), measured from the docs page read | | [Alation AI Agent SDK MCP server](/servers/alation-mcp/) | Remote and local | OAuth client credentials of a service account (role-limited) or user refresh token; remote MCP at /ai/mcp needs client_id and client_secret (no dynamic client registration). | Not stated in the source read | 69 tokens for 4 tool definitions (3 with a description), measured from the docs page read | | [Amazon Redshift MCP Server (AWS Labs)](/servers/awslabs-redshift-mcp-server/) | Local (stdio) | AWS credentials (profile/env) with IAM and Redshift Data API; database user privileges. | Configurable | 238 tokens for 8 tool definitions (8 with a description), measured from the README read | | [Atlan MCP Server](/servers/atlan-mcp/) | Remote (hosted) | OAuth per user (calls run as the signed-in user) or Atlan API key as Bearer token (shared service identity); local server uses API key + base URL. | Configurable | 1,770 tokens for 42 tool definitions (42 with a description), measured from the docs page read | | [BigQuery remote MCP server](/servers/bigquery-remote-mcp/) | Remote (hosted) | OAuth 2.0 with IAM (Google Cloud identities, OAuth client, or agent identity). | Configurable | 1,487 tokens for 9 tool definitions (9 with a description), measured from the docs page read | | [MCP Toolbox for Databases (BigQuery tools)](/servers/mcp-toolbox-bigquery/) | Remote and local | Application Default Credentials, client OAuth token forwarding, or service account impersonation. | Configurable | 156 tokens for 9 tool definitions (9 with a description), measured from the docs page read | | [Collibra MCP Server (chip)](/servers/collibra-mcp/) | Remote and local | Collibra username/password: server-wide credentials over stdio (mcp.yaml or env vars), or client-supplied Basic Auth headers per request over HTTP; tools respect the connecting user's DGC scopes. | Configurable | 2,914 tokens for 41 tool definitions (41 with a description), measured from the README read | | [Cube MCP server](/servers/cube-mcp/) | Remote (hosted) | OAuth Authorization Code + PKCE (scope mcp-agent-access, public client cube-mcp-client), or a Cube API key exchanged for a token (client_id cube-api-key). | Read and write | 1,093 tokens for 30 tool definitions (30 with a description), measured from the docs page read | | [Databricks Labs MCP servers](/servers/databrickslabs-mcp/) | Remote and local | Databricks CLI-style credentials; Unity Catalog permissions. | Not stated in the source read | Not measured: README says the tool list is inferred at startup from the Unity Catalog functions, vector search indexes and Genie spaces the user configures, so no fixed per-tool list or descriptions are published. | | [Databricks managed MCP servers](/servers/databricks-managed-mcp/) | Remote (hosted) | Databricks workspace sign-in; caller's Unity Catalog permissions; Unity Gateway grants and policies. | Configurable | 94 tokens for 4 tool definitions (4 with a description), measured from the docs page read | | [DataHub MCP Server](/servers/datahub-mcp/) | Remote and local | Stdio: DATAHUB_GMS_URL + DATAHUB_GMS_TOKEN (or ~/.datahubenv); HTTP: per-request DataHub personal access token as Authorization Bearer, no shared server token. | Configurable | 567 tokens for 22 tool definitions (22 with a description), measured from the README read | | [dbt MCP Server](/servers/dbt-mcp/) | Remote and local | dbt service/personal token and environment IDs for platform tools; OAuth or token per docs; local dbt project for CLI tools. | Configurable | 1,061 tokens for 63 tool definitions (63 with a description), measured from the README read | | [Euno MCP](/servers/euno-mcp/) | Remote (hosted) | OAuth sign-in (default) or API key and headers; active persona scopes data and actions. | Configurable | 760 tokens for 56 tool definitions (42 with a description), measured from the docs page read | | [Fivetran MCP Server](/servers/fivetran-mcp/) | Remote and local | Fivetran API key and API secret (FIVETRAN_API_KEY, FIVETRAN_API_SECRET) from the Fivetran dashboard; one shared credential, not per-user OAuth. Streamable HTTP mode forwards headers; OAuth resource-server support is scaffolding only. | Configurable | 299 tokens for 53 tool definitions (2 with a description), measured from the README read | | [Hex MCP server](/servers/hex-mcp/) | Remote (hosted) | OAuth authorization to the user's Hex workspace (server URL app.hex.tech/mcp, or the customer's single-tenant, EU or HIPAA domain); requires Explorer role or higher. | Read and write | 252 tokens for 18 tool definitions (18 with a description), measured from the docs page read | | [Looker tools in MCP Toolbox for Databases](/servers/looker-mcp-toolbox/) | Remote and local | Looker API client_id and client_secret in the Looker source config, or use_client_oauth to pass the caller's OAuth token (default false). | Configurable | 1,475 tokens for 55 tool definitions (55 with a description), measured from the docs index read | | [Monte Carlo MCP Server](/servers/monte-carlo-mcp/) | Remote (hosted) | OAuth 2.1 with Dynamic Client Registration (user identity, Editor role or above), or MCP-server-type API key id:secret as Bearer token via mcp-remote. | Configurable | Not measured: no published tool-definition text was read on 10 Nov 2026 | | [Omni MCP Server](/servers/omni-mcp/) | Remote (hosted) | OAuth 2.1 with PKCE (creates an MCP OAuth PAT for the user), or an API key (personal token or Organization API key; Organization keys can act as a user with X-MCP-User-ID). | Configurable | 466 tokens for 27 tool definitions (27 with a description), measured from the docs page read | | [OpenMetadata MCP Server](/servers/openmetadata-mcp/) | Remote (hosted) | OAuth 2.0 via existing SSO (per user), personal access token, or Bot JWT as Authorization Bearer. | Read and write | 1,249 tokens for 15 tool definitions (15 with a description), measured from the docs page read | | [Power BI Modeling MCP Server](/servers/powerbi-modeling-mcp/) | Remote and local | Local: Microsoft Entra interactive sign-in (default) or service principal (client secret or certificate). Hosted: Entra ID as the signed-in user. | Configurable | 387 tokens for 21 tool definitions (21 with a description), measured from the README read | | [Secoda MCP](/servers/secoda-mcp/) | Remote and local | Secoda API token (API_TOKEN env var) for the local FastMCP wrapper; docs also describe a Secoda-hosted MCP endpoint authenticated against workspace and AI permissions. | Not stated in the source read | 63 tokens for 6 tool definitions (6 with a description), measured from the source code read | | [Select Star MCP Server](/servers/select-star-mcp/) | Remote (hosted) | Select Star API token sent as Authorization: Bearer header (mcp-remote proxy used for Claude Desktop). | Read only | 98 tokens for 7 tool definitions (7 with a description), measured from the docs page read | | [Sifflet MCP Server](/servers/sifflet-mcp/) | Local (stdio) | Sifflet API token (Viewer role for most tools, Editor for monitor generation) and backend URL as environment variables. | Read and write | 662 tokens for 10 tool definitions (10 with a description), measured from the source code read | | [Sigma MCP server](/servers/sigma-mcp/) | Remote (hosted) | OAuth (HTTP transport) inheriting the user's Sigma account permissions; API-key access token with scope mcp:access (Beta). | Read and write | Not measured: The docs page describes features (search, explore, analyze, build workbooks, screenshots, agents) in prose with example prompts but publishes no tool names or per-tool list. | | [mcp-snowflake-server (community)](/servers/isaacwasserman-mcp-snowflake-server/) | Local (stdio) | Snowflake user credentials (password, private key, external browser) or connections.toml. | Configurable | 103 tokens for 8 tool definitions (8 with a description), measured from the README read | | [Snowflake Labs MCP Server](/servers/snowflake-labs-mcp/) | Remote and local | Snowflake Python connector: password/PAT, key pair, OAuth, SSO, MFA; honors the connection role's RBAC. | Configurable | Not measured: The Snowflake Labs server, marked deprecated in its README, which describes config-driven tool groups (Cortex Search, Analyst, Agent, object management, SQL execution, semantic views) but publishes no per-tool names or descriptions; facts.json lists these six groups. | | [Snowflake-managed MCP server](/servers/snowflake-managed-mcp/) | Remote (hosted) | Snowflake OAuth (default) or External OAuth; PAT supported; RBAC per MCP server and per tool. | Configurable | Not measured: The docs page defines user-configured tool types (CORTEX_AGENT_RUN, CODING_AGENT, CORTEX_SEARCH_SERVICE_QUERY, CORTEX_ANALYST_MESSAGE, SYSTEM_EXECUTE_SQL, GENERIC) used to build tools in a CREATE MCP SERVER specification; it publishes no fixed list of named tools, so facts.json lists tool types not tools. | | [SYNQ MCP (Coalesce Quality)](/servers/synq-mcp/) | Remote (hosted) | OAuth 2 with PKCE per user with selectable permission tier (read, write, governance, deploy); legacy workspace token (st-...) as Bearer. | Configurable | 4,550 tokens for 134 tool definitions (134 with a description), measured from the docs page read | | [Tableau MCP](/servers/tableau-mcp/) | Remote and local | Hosted: OAuth 2.1 with each user's own Tableau Cloud identity. Self-hosted: PAT (default), direct-trust (Connected Apps), OAuth, passthrough, or unified access tokens. | Configurable | Not measured: the source read on 10 Nov 2026 holds the definitions of 2 of the roughly 40 Tableau MCP tools (get-datasource-metadata and query-datasource), and one of those descriptions is imported from a file that was not read, so a measurement would cover a small part of the server. | | [ThoughtSpot MCP Server](/servers/thoughtspot-mcp-server/) | Remote (hosted) | OAuth to a ThoughtSpot instance (dynamic client registration, or manually registered client id and secret); bearer-token and /sse paths also exist; the user chooses or is bound to a ThoughtSpot instance URL. | Read and write | at least 2,149 tokens for 20 tool definitions (20 with a description), measured from the source code read | Filed by data tool, read 10 Nov 2026. Scope: warehouses, transformation, ingestion connectors, bi and analytics, catalogs and metadata, data observability. ## What each server documents on clients, auth and scoping - [Airbyte MCP (hosted)](/servers/airbyte-mcp/): The agent acts on behalf of the Airbyte user who signed in, and every tool runs with that user's Airbyte permissions. Direct reads use the credentials stored in the connector rather than the asking user, so source-side access is not per-user. With the context layer enabled, agents can read live source data, query supported destinations and search synced data by keyword, meaning or both. - [Alation AI Agent SDK MCP server](/servers/alation-mcp/): The SDK and its MCP server let AI agents use Alation catalog metadata for curation, search and discovery, and data analyst agents. The remote MCP server lists tools dynamically, filtering out tools the current user cannot use. Service-account authentication gives the agent the data and features allowed by the role assigned to the service account. SQL tools need per-data-product data warehouse credentials configured once. - [Amazon Redshift MCP Server (AWS Labs)](/servers/awslabs-redshift-mcp-server/): The README lists 8 tools for discovery, query and cluster review, scoped to the AWS credentials in use. - [Atlan MCP Server](/servers/atlan-mcp/): The MCP server serves the catalog as a context layer to AI clients (Claude, Cursor, ChatGPT, Gemini, Copilot) and automation platforms. OAuth makes each tool call run as the signed-in user, so tools run with their own permissions and actions are attributed to them. Guests and members are scoped to their persona; MCP narrows but never widens what a user's login permits. The tools page lists 41 tools across nine categories. - [BigQuery remote MCP server](/servers/bigquery-remote-mcp/): The server can run queries, get metadata and list resources for AI apps including Claude and ChatGPT. OAuth 2.0 with IAM is used so access follows the authenticated Google Cloud identity. - [MCP Toolbox for Databases (BigQuery tools)](/servers/mcp-toolbox-bigquery/): Tools include ask_data_insights (Conversational Analytics API), forecast and analyze_contribution. With useClientOAuth the client's OAuth token is forwarded so calls run as the asking user. - [Collibra MCP Server (chip)](/servers/collibra-mcp/): The server is described as giving AI agents access to Collibra Data Governance Center for asset discovery, glossary queries and asset details. Permission scoping follows the connecting Collibra user: tools needing extra DGC scopes are flagged, and a missing scope returns a permission error. Natural-language discovery tools use Collibra AI and consume Collibra Units. With the HTTP transport, MCP clients are advised to supply their own Basic Auth headers per request; with stdio, one server-wide credential is used. - [Cube MCP server](/servers/cube-mcp/): Every tool runs as the authenticated user, and queries respect Cube permissions including row-level security. With an API key, a token can act as the key's owner, as another user, or be limited to specific tools. Which tools a user sees depends on their role; Viewer role or higher is needed, and data model editing tools show only for roles allowed to edit the model (Admin and Developer by default). - [Databricks Labs MCP servers](/servers/databrickslabs-mcp/): Tools are inferred at startup from the functions and vector search indexes in a given Unity Catalog schema plus listed Genie spaces. - [Databricks managed MCP servers](/servers/databricks-managed-mcp/): System.ai.genie_one_mcp answers natural-language analytics questions grounded in Genie Ontology. The dbsql MCP runs SQL using the caller's Unity Catalog and warehouse permissions, so results are scoped to the asking user. - [DataHub MCP Server](/servers/datahub-mcp/): The server is built to serve DataHub context to AI agents: search, lineage, query history and entity metadata. In the Docker HTTP deployment each MCP client supplies its own DataHub token, so DataHub permissions and audit identity are preserved per user. An optional get_me tool (TOOLS_IS_USER_ENABLED=true) returns the authenticated user's profile and group memberships. - [dbt MCP Server](/servers/dbt-mcp/): The dbt MCP server provides AI agents with context of a project in dbt Core, dbt Fusion and dbt Platform. - [Euno MCP](/servers/euno-mcp/): Euno MCP uses the active persona; if none can be identified the user is asked to choose. Euno MCP works with Cursor, Claude, ChatGPT in Developer Mode and Claude Code. Euno MCP has a dedicated Gemini Enterprise setup path besides direct, bridge and API-key setups. - [Fivetran MCP Server](/servers/fivetran-mcp/): The server uses one Fivetran API key and secret, so agent actions run with that credential's scope instead of the asking user's identity. The README states OAuth resource-server support in streamable-http mode is not yet functional. Two discovery tools plus one execution tool per allowed resource/action pair keep the tool list small and filtered by scope. - [Hex MCP server](/servers/hex-mcp/): Threads created through the MCP server behave like Threads in the Hex app and select among the data connections the user can access. The MCP server enforces workspace roles and project permissions; editing a cell needs Can Edit or Full Access on that project, even for Admins. Knowledge tools need the Explorer role or above and project editing tools need Editor or above. - [Looker tools in MCP Toolbox for Databases](/servers/looker-mcp-toolbox/): The Looker source can pass the caller's OAuth token instead of one service credential via use_client_oauth, which defaults to false. The default source config authenticates with a single client_id and client_secret, so identity is that API user unless OAuth is enabled. The source can hide models, explores and fields from agents with show_hidden_models, show_hidden_explores and show_hidden_fields (all default true). - [Monte Carlo MCP Server](/servers/monte-carlo-mcp/): Monte Carlo says connected agents can investigate alerts, explore assets and lineage, and create and manage monitors. OAuth lets any HTTP-transport MCP client connect with no API keys, and the connector is listed in the Claude Connectors Directory. An Editor role or above is required to use the server. An agent_observability toolset adds get_agent_metadata, get_agent_conversation and get_agent_trace for inspecting AI agents. - [Omni MCP Server](/servers/omni-mcp/): MCP tools run with the permissions of a specific Omni user, so results include only what that user can access. With OAuth, all MCP queries use the user's Omni permissions and access controls. An Organization API key can run tools as another user via the X-MCP-User-ID header, and X-MCP-Model-ID can pin a model. - [OpenMetadata MCP Server](/servers/openmetadata-mcp/): Get_user_context returns the authenticated caller's identity, roles, teams, domains and active persona, with no parameter to look up another user. A personal access token gives the agent the same role and access policy as the user. OAuth 2.0 runs tool calls with the user's existing SSO login (Google, Azure, Okta, Auth0, LDAP, SAML). - [Power BI Modeling MCP Server](/servers/powerbi-modeling-mcp/): The hosted server prompts for Microsoft Entra sign-in on first tool call and then acts with the user's permissions. Fabric IQ, Microsoft's consumption server, answers questions over semantic models and respects permissions, row-level security and object-level security. Microsoft states that clients operate under the user's Fabric RBAC permissions. - [Secoda MCP](/servers/secoda-mcp/): The Secoda-hosted MCP server lets AI tools search, query SQL, read lineage and glossary terms. Authentication verifies the user's workspace and AI permissions, and operations respect governance and access controls. The open-source wrapper authenticates with a single API_TOKEN, i.e. one shared token per server process. - [Select Star MCP Server](/servers/select-star-mcp/): The MCP server is meant to bring catalog metadata and lineage to Cursor, Claude Desktop and VS Code GitHub Copilot. Authentication is a Select Star API token sent as a Bearer header; no per-user OAuth is described, and the docs do not say results are scoped to the asking user. None found: A statement that Select Star MCP results are limited to the asking user's own permissions. - [Sifflet MCP Server](/servers/sifflet-mcp/): The README documents setup for Cursor and Claude Desktop through uvx sifflet-mcp@latest. - [Sigma MCP server](/servers/sigma-mcp/): With OAuth the server inherits the user's Sigma account permissions. Access needs the 'Use Sigma MCP with OAuth' account-type permission plus at least Can view on documents or Can use on connections. Agents can call Sigma agents defined in workbooks if the user has Can view on the workbook. - [mcp-snowflake-server (community)](/servers/isaacwasserman-mcp-snowflake-server/): The server exposes a memo://insights resource and an append_insight tool that stores agent-discovered insights. When prefetch is enabled it exposes per-table schema summaries as resources. - [Snowflake Labs MCP Server](/servers/snowflake-labs-mcp/): The server served Cortex Search, Cortex Analyst and Cortex Agent services to MCP clients as tools, configured in a YAML file. The server honors the RBAC grants of the role used for the connection, so access is scoped to that Snowflake role. - [Snowflake-managed MCP server](/servers/snowflake-managed-mcp/): MCP clients discover and invoke Cortex Analyst, Cortex Search and Cortex Agents as tools on the managed server. Access to the MCP server does not give access to tools; privileges are granted per tool, so results depend on the calling role. - [SYNQ MCP (Coalesce Quality)](/servers/synq-mcp/): OAuth 2 with PKCE means actions are recorded as the real user with a via MCP badge. The set of tools an assistant sees depends on the permissions the user grants at consent, and the consent screen offers only what the user's role allows. The iam tool reports the caller's workspace, granted scopes and user. - [Tableau MCP](/servers/tableau-mcp/): The hosted Tableau MCP at mcp.tableau.com uses OAuth 2.1 so each user signs in as themselves and per-user Tableau permissions are enforced. Authentication to the Tableau REST APIs is selectable between pat (default), direct-trust, oauth and passthrough. The Tableau docs state the authenticated user must have access to the published data sources the agent will use. - [ThoughtSpot MCP Server](/servers/thoughtspot-mcp-server/): Spotter 'handles authentication securely, and returns live answers from your own data' per the README. The model tools are offered as a group and gated on the user's data-modeling privileges, and get_data is gated on the data-download privilege at runtime. OAuth client registration can be tied to a ThoughtSpot instance so users do not enter the instance URL. **What Euno's server does not do.** Euno's server gives an agent governed context about the data; it does not run the query, which is the job of the warehouse's own server. ## Other situations - [A data platform lead whose analysts no longer trust the dashboards](/situations/untrusted-dashboards/) - [An analytics engineering team on dbt and Snowflake or Databricks](/situations/dbt-warehouse/) - [A governance lead who must show where a regulated field flows](/situations/regulated-field/) - [A BI team with thousands of reports and no map of what feeds them](/situations/bi-sprawl/) ## Sources - docs.airbyte.com/platform/airbyte-mcp/tools.md: https://docs.airbyte.com/platform/airbyte-mcp/tools.md (read 10 Nov 2026) - docs.airbyte.com/platform/airbyte-mcp.md: https://docs.airbyte.com/platform/airbyte-mcp.md (read 10 Nov 2026) - github.com/Alation/alation-ai-agent-sdk/blob/main/README.md: https://github.com/Alation/alation-ai-agent-sdk/blob/main/README.md (read 10 Nov 2026) - github.com/Alation/alation-ai-agent-sdk/blob/main/guides/authentication.md: https://github.com/Alation/alation-ai-agent-sdk/blob/main/guides/authentication.md (read 10 Nov 2026) - raw.githubusercontent.com/awslabs/mcp/main/src/redshift-mcp-server/README.md: https://raw.githubusercontent.com/awslabs/mcp/main/src/redshift-mcp-server/README.md (read 10 Nov 2026) - docs.atlan.com/product/capabilities/atlan-ai/how-tos/remote-mcp-overview: https://docs.atlan.com/product/capabilities/atlan-ai/how-tos/remote-mcp-overview (read 10 Nov 2026) - docs.atlan.com/product/capabilities/atlan-ai/references/mcp-security: https://docs.atlan.com/product/capabilities/atlan-ai/references/mcp-security (read 10 Nov 2026) - docs.atlan.com/product/capabilities/atlan-ai/references/mcp-tools: https://docs.atlan.com/product/capabilities/atlan-ai/references/mcp-tools (read 10 Nov 2026) - docs.cloud.google.com/bigquery/docs/use-bigquery-mcp: https://docs.cloud.google.com/bigquery/docs/use-bigquery-mcp (read 10 Nov 2026) - raw.githubusercontent.com/googleapis/mcp-toolbox/main/docs/en/integrations/bigquery/prebuilt-configs/bigquery.md: https://raw.githubusercontent.com/googleapis/mcp-toolbox/main/docs/en/integrations/bigquery/prebuilt-configs/bigquery.md (read 10 Nov 2026) - raw.githubusercontent.com/googleapis/mcp-toolbox/main/docs/en/integrations/bigquery/source.md: https://raw.githubusercontent.com/googleapis/mcp-toolbox/main/docs/en/integrations/bigquery/source.md (read 10 Nov 2026) - github.com/collibra/chip/blob/main/README.md: https://github.com/collibra/chip/blob/main/README.md (read 10 Nov 2026) - docs.cube.dev/docs/integrations/mcp-server.md: https://docs.cube.dev/docs/integrations/mcp-server.md (read 10 Nov 2026) - raw.githubusercontent.com/databrickslabs/mcp/master/README.md: https://raw.githubusercontent.com/databrickslabs/mcp/master/README.md (read 10 Nov 2026) - docs.databricks.com/aws/en/generative-ai/mcp/managed-mcp: https://docs.databricks.com/aws/en/generative-ai/mcp/managed-mcp (read 10 Nov 2026) - raw.githubusercontent.com/acryldata/mcp-server-datahub/main/README.md: https://raw.githubusercontent.com/acryldata/mcp-server-datahub/main/README.md (read 10 Nov 2026) - raw.githubusercontent.com/dbt-labs/dbt-mcp/main/README.md: https://raw.githubusercontent.com/dbt-labs/dbt-mcp/main/README.md (read 10 Nov 2026) - docs.euno.ai/ai/mcp-setup.md: https://docs.euno.ai/ai/mcp-setup.md (read 10 Nov 2026) - raw.githubusercontent.com/fivetran/fivetran-mcp/main/README.md: https://raw.githubusercontent.com/fivetran/fivetran-mcp/main/README.md (read 10 Nov 2026) - learn.hex.tech/docs/api-integrations/mcp-server: https://learn.hex.tech/docs/api-integrations/mcp-server (read 10 Nov 2026) - mcp-toolbox.dev/integrations/looker/source: https://mcp-toolbox.dev/integrations/looker/source/ (read 10 Nov 2026) - docs.getmontecarlo.com/docs/mcp-server: https://docs.getmontecarlo.com/docs/mcp-server (read 10 Nov 2026) - docs.getmontecarlo.com/docs/mcp-server-reference-and-operations: https://docs.getmontecarlo.com/docs/mcp-server-reference-and-operations (read 10 Nov 2026) - docs.omni.co/ai/mcp/tools.md: https://docs.omni.co/ai/mcp/tools.md (read 10 Nov 2026) - docs.omni.co/ai/mcp/authentication.md: https://docs.omni.co/ai/mcp/authentication.md (read 10 Nov 2026) - docs.open-metadata.org/v2.0.x/how-to-guides/mcp/reference.md: https://docs.open-metadata.org/v2.0.x/how-to-guides/mcp/reference.md (read 10 Nov 2026) - docs.open-metadata.org/v2.0.x/how-to-guides/mcp.md: https://docs.open-metadata.org/v2.0.x/how-to-guides/mcp.md (read 10 Nov 2026) - learn.microsoft.com/en-us/power-bi/developer/mcp/power-bi-authoring-mcp: https://learn.microsoft.com/en-us/power-bi/developer/mcp/power-bi-authoring-mcp (read 10 Nov 2026) - learn.microsoft.com/en-us/power-bi/developer/mcp/mcp-servers-overview: https://learn.microsoft.com/en-us/power-bi/developer/mcp/mcp-servers-overview (read 10 Nov 2026) - raw.githubusercontent.com/microsoft/powerbi-modeling-mcp/main/README.md: https://raw.githubusercontent.com/microsoft/powerbi-modeling-mcp/main/README.md (read 10 Nov 2026) - docs.secoda.co/features/ai-assistant/secoda-mcp-server.md: https://docs.secoda.co/features/ai-assistant/secoda-mcp-server.md (read 10 Nov 2026) - selectstar.com/product/mcp-for-data: https://selectstar.com/product/mcp-for-data (read 10 Nov 2026) - docs.selectstar.com/features/mcp-server: https://docs.selectstar.com/features/mcp-server (read 10 Nov 2026) - github.com/siffletdata/sifflet-mcp/blob/main/README.md: https://github.com/siffletdata/sifflet-mcp/blob/main/README.md (read 10 Nov 2026) - help.sigmacomputing.com/docs/use-sigma-mcp-server: https://help.sigmacomputing.com/docs/use-sigma-mcp-server (read 10 Nov 2026) - raw.githubusercontent.com/isaacwasserman/mcp-snowflake-server/main/README.md: https://raw.githubusercontent.com/isaacwasserman/mcp-snowflake-server/main/README.md (read 10 Nov 2026) - raw.githubusercontent.com/Snowflake-Labs/mcp/main/README.md: https://raw.githubusercontent.com/Snowflake-Labs/mcp/main/README.md (read 10 Nov 2026) - docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-mcp: https://docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-mcp (read 10 Nov 2026) - docs.synq.io/scout/mcp.md: https://docs.synq.io/scout/mcp.md (read 10 Nov 2026) - github.com/tableau/tableau-mcp: https://github.com/tableau/tableau-mcp (read 10 Nov 2026) - tableau.github.io/tableau-mcp/docs/configuration/mcp-config/env-vars: https://tableau.github.io/tableau-mcp/docs/configuration/mcp-config/env-vars (read 10 Nov 2026) - tableau.github.io/tableau-mcp/docs/configuration/mcp-config/authentication: https://tableau.github.io/tableau-mcp/docs/configuration/mcp-config/authentication (read 10 Nov 2026) - raw.githubusercontent.com/thoughtspot/mcp-server/main/README.md: https://raw.githubusercontent.com/thoughtspot/mcp-server/main/README.md (read 10 Nov 2026) - raw.githubusercontent.com/thoughtspot/mcp-server/main/src/servers/tool-definitions.ts: https://raw.githubusercontent.com/thoughtspot/mcp-server/main/src/servers/tool-definitions.ts (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/situations/agents-this-year/ --- # To show where a regulated field flows, check lineage depth, classification, scoping and audit in each server SITUATION, read 10 Nov 2026 **The short answer.** A governance lead who must show where a regulated field flows needs column-level lineage, classification tags and an audit trail of what the agent read. Twelve catalog and warehouse servers are in scope; the checks below point to the fields and facts that answer each, read 10 Nov 2026. Tracing a regulated field means following one column through transformations into reports, and knowing who asked. Catalog servers carry the lineage and the tags; warehouse servers carry the grants that limit what an agent sees. Where a server's documentation says nothing about audit, the entry says none found on 10 Nov 2026. ## What to check | Check | The field or fact that answers it | | --- | --- | | Does lineage reach the column? | Lineage and impact facts on each entry | | Can the agent read classifications and tags? | Definitions, ownership and rules facts | | Is each call scoped to the person asking? | Auth field | | Is there an audit trail? | Access and safety facts | ## The published facts, for the 16 servers in scope | Server | Kind | Auth | Access | Transport | | --- | --- | --- | --- | --- | | [Alation AI Agent SDK MCP server](/servers/alation-mcp/) | Official | OAuth client credentials of a service account (role-limited) or user refresh token; remote MCP at /ai/mcp needs client_id and client_secret (no dynamic client registration). | Not stated in the source read | Remote and local | | [Amazon Redshift MCP Server (AWS Labs)](/servers/awslabs-redshift-mcp-server/) | Official, Labs org | AWS credentials (profile/env) with IAM and Redshift Data API; database user privileges. | Configurable | Local (stdio) | | [Atlan MCP Server](/servers/atlan-mcp/) | Official | OAuth per user (calls run as the signed-in user) or Atlan API key as Bearer token (shared service identity); local server uses API key + base URL. | Configurable | Remote (hosted) | | [BigQuery remote MCP server](/servers/bigquery-remote-mcp/) | Official | OAuth 2.0 with IAM (Google Cloud identities, OAuth client, or agent identity). | Configurable | Remote (hosted) | | [MCP Toolbox for Databases (BigQuery tools)](/servers/mcp-toolbox-bigquery/) | Official | Application Default Credentials, client OAuth token forwarding, or service account impersonation. | Configurable | Remote and local | | [Collibra MCP Server (chip)](/servers/collibra-mcp/) | Official | Collibra username/password: server-wide credentials over stdio (mcp.yaml or env vars), or client-supplied Basic Auth headers per request over HTTP; tools respect the connecting user's DGC scopes. | Configurable | Remote and local | | [Databricks Labs MCP servers](/servers/databrickslabs-mcp/) | Official, Labs org | Databricks CLI-style credentials; Unity Catalog permissions. | Not stated in the source read | Remote and local | | [Databricks managed MCP servers](/servers/databricks-managed-mcp/) | Official | Databricks workspace sign-in; caller's Unity Catalog permissions; Unity Gateway grants and policies. | Configurable | Remote (hosted) | | [DataHub MCP Server](/servers/datahub-mcp/) | Official | Stdio: DATAHUB_GMS_URL + DATAHUB_GMS_TOKEN (or ~/.datahubenv); HTTP: per-request DataHub personal access token as Authorization Bearer, no shared server token. | Configurable | Remote and local | | [Euno MCP](/servers/euno-mcp/) | Official | OAuth sign-in (default) or API key and headers; active persona scopes data and actions. | Configurable | Remote (hosted) | | [OpenMetadata MCP Server](/servers/openmetadata-mcp/) | Official | OAuth 2.0 via existing SSO (per user), personal access token, or Bot JWT as Authorization Bearer. | Read and write | Remote (hosted) | | [Secoda MCP](/servers/secoda-mcp/) | Official | Secoda API token (API_TOKEN env var) for the local FastMCP wrapper; docs also describe a Secoda-hosted MCP endpoint authenticated against workspace and AI permissions. | Not stated in the source read | Remote and local | | [Select Star MCP Server](/servers/select-star-mcp/) | Official | Select Star API token sent as Authorization: Bearer header (mcp-remote proxy used for Claude Desktop). | Read only | Remote (hosted) | | [mcp-snowflake-server (community)](/servers/isaacwasserman-mcp-snowflake-server/) | Community | Snowflake user credentials (password, private key, external browser) or connections.toml. | Configurable | Local (stdio) | | [Snowflake Labs MCP Server](/servers/snowflake-labs-mcp/) | Official, Labs org | Snowflake Python connector: password/PAT, key pair, OAuth, SSO, MFA; honors the connection role's RBAC. | Configurable | Remote and local | | [Snowflake-managed MCP server](/servers/snowflake-managed-mcp/) | Official | Snowflake OAuth (default) or External OAuth; PAT supported; RBAC per MCP server and per tool. | Configurable | Remote (hosted) | Filed by data tool, read 10 Nov 2026. Scope: catalogs and metadata, warehouses. ## What each server documents on access and safety - [Alation AI Agent SDK MCP server](/servers/alation-mcp/): Metadata-writing tools update_catalog_asset_metadata and check_job_status were removed from the SDK (Jan 2026), leaving curation to the REST API. The MCP server supports an explicit enabled-tools list (CLI argument or ALATION_ENABLED_TOOLS) to limit the tools exposed. The remote MCP server does not support dynamic client registration; clients must supply a client_id and client_secret. - [Amazon Redshift MCP Server (AWS Labs)](/servers/awslabs-redshift-mcp-server/): The server is read-only by default; read-write mode must be selected with ACCESS_MODE. The database user's privileges are the security boundary, not the server. In read-write mode each write prompts for confirmation via MCP elicitation unless UNSAFE_SKIP_WRITE_CONFIRMATION is set. Authentication uses AWS credentials via AWS CLI, profile or environment variables, with IAM permissions for the Data API. - [Atlan MCP Server](/servers/atlan-mcp/): An optional read-only mode removes every write, admin and lifecycle tool while keeping search and lineage; it is requested from Atlan support. Tools that change anything propose first and wait for approval before writing. Every tool call is logged as a structured event with tool name, client, tenant, request ID and status. Tools that execute SQL route it through a read-only validator that permits only SELECT-class queries. - [BigQuery remote MCP server](/servers/bigquery-remote-mcp/): Execute_sql_readonly is restricted to SELECT statements and is preferred over execute_sql. Execute_sql can run INSERT, UPDATE, DELETE and CREATE statements. Google remote MCP servers offer centralized audit logging and optional Model Armor protection. - [MCP Toolbox for Databases (BigQuery tools)](/servers/mcp-toolbox-bigquery/): Source options readOnly and writeMode (allowed, blocked, protected) enforce read-only operation; blocked allows only SELECT. AllowedDatasets restricts tools to listed datasets and impersonateServiceAccount supports service account impersonation. BIGQUERY_MAXIMUM_BYTES_BILLED caps per-query bytes scanned. - [Collibra MCP Server (chip)](/servers/collibra-mcp/): Data quality tools (18 tools, several that write or delete) are off by default and gated behind a --data-quality flag. Specific tools can be enabled or disabled through flags, environment variables or mcp.yaml; by default all are enabled. Destructive DQ tools use a confirm checkpoint (confirm=false by default is read-only). HTTP mode binds to localhost only, and credentials should be kept in environment variables. - [Databricks Labs MCP servers](/servers/databrickslabs-mcp/): Databricks managed MCP note in the README states permissions are always enforced for managed servers. - [Databricks managed MCP servers](/servers/databricks-managed-mcp/): Each system.ai MCP is a Unity Catalog securable governed by grants and per-call policies via Unity Gateway. Genie Agent MCP calls bypass Unity Gateway and rely on permissions of the Genie Agent and its data. - [DataHub MCP Server](/servers/datahub-mcp/): Mutation tools are disabled by default and need TOOLS_IS_MUTATION_ENABLED=true, so the default is read-only. The HTTP entry point refuses to start if a shared server-wide token is configured, forcing per-user tokens. Tokens are accepted only in the Authorization header; query-string tokens are rejected. - [Euno MCP](/servers/euno-mcp/): Euno MCP includes read and write tools; write tools such as catalog creates and benchmark changes ask for client confirmation. Euno MCP catalog reads are published read-only and catalog writes are marked destructive; deletes need confirm=true. Euno enforces authentication and persona permissions on the server, independent of client confirmation. Context benchmark tools are accepted only for an account admin persona. - [OpenMetadata MCP Server](/servers/openmetadata-mcp/): Every tool is labeled Read or Write in the reference. Unattended agents should use a Bot service account with a role-based access policy, kept separate from personal logins and auditable. The transport is Streamable HTTP (MCP spec 26 Mar 2025). - [Secoda MCP](/servers/secoda-mcp/): The tools shown are search, SQL execution, entity retrieval, lineage and glossary; run_sql can execute SQL, and the docs do not describe a read-only mode. Admin must enable MCP and the user needs the right Secoda permissions. - [Select Star MCP Server](/servers/select-star-mcp/): Seven tools are listed, all search/get/traverse operations; no write tools appear on the page. Troubleshooting says the token must be formatted with a Bearer prefix. - [mcp-snowflake-server (community)](/servers/isaacwasserman-mcp-snowflake-server/): Write operations are disabled by default and enabled only with --allow-write. Tool exclusion and database/schema/table exclusion patterns are supported. Authentication uses Snowflake user credentials, a private key path or external browser; the README shows plain passwords in config. - [Snowflake Labs MCP Server](/servers/snowflake-labs-mcp/): SQL statement types are allowed or denied per type in a sql_statement_permissions config block. Authentication supports password, key pair, OAuth, SSO and MFA via the Snowflake connector. - [Snowflake-managed MCP server](/servers/snowflake-managed-mcp/): Authentication is Snowflake OAuth by default, with optional External OAuth, and role-based access control on the server and tools. Docs list distinct privileges: USAGE on the MCP server plus SELECT on the semantic view, USAGE on the Cortex Search service, agent, or UDF. Docs warn that tool annotations are client hints and not security controls; RBAC is the control. The managed server is available in government regions except Azure US Gov Virginia and has not completed formal FedRAMP assessment. **What Euno's server does not do.** Euno's server gives an agent governed context about the data; it does not run the query, which is the job of the warehouse's own server. ## Other situations - [A data platform lead whose analysts no longer trust the dashboards](/situations/untrusted-dashboards/) - [An analytics engineering team on dbt and Snowflake or Databricks](/situations/dbt-warehouse/) - [A company putting AI agents on its data this year](/situations/agents-this-year/) - [A BI team with thousands of reports and no map of what feeds them](/situations/bi-sprawl/) ## Sources - github.com/Alation/alation-ai-agent-sdk/blob/main/README.md: https://github.com/Alation/alation-ai-agent-sdk/blob/main/README.md (read 10 Nov 2026) - raw.githubusercontent.com/awslabs/mcp/main/src/redshift-mcp-server/README.md: https://raw.githubusercontent.com/awslabs/mcp/main/src/redshift-mcp-server/README.md (read 10 Nov 2026) - docs.atlan.com/product/capabilities/atlan-ai/references/mcp-security: https://docs.atlan.com/product/capabilities/atlan-ai/references/mcp-security (read 10 Nov 2026) - docs.atlan.com/product/capabilities/atlan-ai/references/mcp-tools: https://docs.atlan.com/product/capabilities/atlan-ai/references/mcp-tools (read 10 Nov 2026) - docs.cloud.google.com/bigquery/docs/reference/mcp: https://docs.cloud.google.com/bigquery/docs/reference/mcp (read 10 Nov 2026) - docs.cloud.google.com/bigquery/docs/use-bigquery-mcp: https://docs.cloud.google.com/bigquery/docs/use-bigquery-mcp (read 10 Nov 2026) - raw.githubusercontent.com/googleapis/mcp-toolbox/main/docs/en/integrations/bigquery/source.md: https://raw.githubusercontent.com/googleapis/mcp-toolbox/main/docs/en/integrations/bigquery/source.md (read 10 Nov 2026) - raw.githubusercontent.com/googleapis/mcp-toolbox/main/docs/en/integrations/bigquery/prebuilt-configs/bigquery.md: https://raw.githubusercontent.com/googleapis/mcp-toolbox/main/docs/en/integrations/bigquery/prebuilt-configs/bigquery.md (read 10 Nov 2026) - github.com/collibra/chip/blob/main/README.md: https://github.com/collibra/chip/blob/main/README.md (read 10 Nov 2026) - raw.githubusercontent.com/databrickslabs/mcp/master/README.md: https://raw.githubusercontent.com/databrickslabs/mcp/master/README.md (read 10 Nov 2026) - docs.databricks.com/aws/en/generative-ai/mcp/managed-mcp: https://docs.databricks.com/aws/en/generative-ai/mcp/managed-mcp (read 10 Nov 2026) - raw.githubusercontent.com/acryldata/mcp-server-datahub/main/README.md: https://raw.githubusercontent.com/acryldata/mcp-server-datahub/main/README.md (read 10 Nov 2026) - docs.euno.ai/ai/mcp-setup.md: https://docs.euno.ai/ai/mcp-setup.md (read 10 Nov 2026) - docs.open-metadata.org/v2.0.x/how-to-guides/mcp/reference.md: https://docs.open-metadata.org/v2.0.x/how-to-guides/mcp/reference.md (read 10 Nov 2026) - docs.open-metadata.org/v2.0.x/how-to-guides/mcp.md: https://docs.open-metadata.org/v2.0.x/how-to-guides/mcp.md (read 10 Nov 2026) - docs.open-metadata.org/v2.0.x/how-to-guides/mcp/connect.md: https://docs.open-metadata.org/v2.0.x/how-to-guides/mcp/connect.md (read 10 Nov 2026) - docs.secoda.co/features/ai-assistant/secoda-mcp-server.md: https://docs.secoda.co/features/ai-assistant/secoda-mcp-server.md (read 10 Nov 2026) - docs.selectstar.com/features/mcp-server: https://docs.selectstar.com/features/mcp-server (read 10 Nov 2026) - raw.githubusercontent.com/isaacwasserman/mcp-snowflake-server/main/README.md: https://raw.githubusercontent.com/isaacwasserman/mcp-snowflake-server/main/README.md (read 10 Nov 2026) - raw.githubusercontent.com/Snowflake-Labs/mcp/main/README.md: https://raw.githubusercontent.com/Snowflake-Labs/mcp/main/README.md (read 10 Nov 2026) - docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-mcp: https://docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-mcp (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/situations/regulated-field/ --- # With thousands of reports and no map, check what each BI and catalog server exposes about what feeds them SITUATION, read 10 Nov 2026 **The short answer.** A BI team with thousands of reports and no map of their sources needs servers that expose data source metadata, semantic models and lineage into reports. Sixteen BI and catalog servers are in scope; the checks below point to the fields that answer each, read 10 Nov 2026, and name no choice. BI servers describe their own content: workbooks, data sources, semantic models, calculated fields. Catalog servers describe how those connect to the warehouse. A map of what feeds each report usually needs both. Several BI servers can also create or edit content; the Access field shows which document a read-only switch. ## What to check | Check | The field or fact that answers it | | --- | --- | | Does the BI server expose data source and calculation metadata? | What it reaches and Definitions facts; tool names | | Can the agent see which tables feed a report? | Lineage and impact facts, on BI and catalog entries | | Can the agent edit reports? | Access field and write tool names | | Is it hosted or local? | Transport field | ## The published facts, for the 16 servers in scope | Server | Kind | Transport | Access | Tools listed | | --- | --- | --- | --- | --- | | [Alation AI Agent SDK MCP server](/servers/alation-mcp/) | Official | Remote and local | Not stated in the source read | 12 named in the evidence | | [Atlan MCP Server](/servers/atlan-mcp/) | Official | Remote (hosted) | Configurable | 41 named in the evidence | | [Collibra MCP Server (chip)](/servers/collibra-mcp/) | Official | Remote and local | Configurable | 49 named in the evidence | | [Cube MCP server](/servers/cube-mcp/) | Official | Remote (hosted) | Read and write | 30 named in the evidence | | [DataHub MCP Server](/servers/datahub-mcp/) | Official | Remote and local | Configurable | 22 named in the evidence | | [Euno MCP](/servers/euno-mcp/) | Official | Remote (hosted) | Configurable | 56 named in the evidence | | [Hex MCP server](/servers/hex-mcp/) | Official | Remote (hosted) | Read and write | 18 named in the evidence | | [Looker tools in MCP Toolbox for Databases](/servers/looker-mcp-toolbox/) | Official | Remote and local | Configurable | 55 named in the evidence | | [Omni MCP Server](/servers/omni-mcp/) | Official | Remote (hosted) | Configurable | 26 named in the evidence | | [OpenMetadata MCP Server](/servers/openmetadata-mcp/) | Official | Remote (hosted) | Read and write | 15 named in the evidence | | [Power BI Modeling MCP Server](/servers/powerbi-modeling-mcp/) | Official | Remote and local | Configurable | 21 named in the evidence | | [Secoda MCP](/servers/secoda-mcp/) | Official | Remote and local | Not stated in the source read | 8 named in the evidence | | [Select Star MCP Server](/servers/select-star-mcp/) | Official | Remote (hosted) | Read only | 7 named in the evidence | | [Sigma MCP server](/servers/sigma-mcp/) | Official | Remote (hosted) | Read and write | 0 named in the evidence | | [Tableau MCP](/servers/tableau-mcp/) | Official | Remote and local | Configurable | 40 named in the evidence | | [ThoughtSpot MCP Server](/servers/thoughtspot-mcp-server/) | Official | Remote (hosted) | Read and write | 20 named in the evidence | Filed by data tool, read 10 Nov 2026. Scope: bi and analytics, catalogs and metadata. ## What each server documents on what it reaches - [Alation AI Agent SDK MCP server](/servers/alation-mcp/): None found: A list of warehouses, transformation or BI tools in the Alation agent SDK README (coverage follows the Alation catalog). - [Atlan MCP Server](/servers/atlan-mcp/): The docs list supported AI clients: Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, ChatGPT, Snowflake, Google ADK, n8n, Microsoft 365 Copilot, Glean and Databricks. Data-source coverage depends on Atlan connectors, not stated on this page. - [Collibra MCP Server (chip)](/servers/collibra-mcp/): None found: A list of warehouses, BI tools or transformation tools supported by the Collibra MCP README (coverage depends on what is catalogued in the Collibra instance). - [Cube MCP server](/servers/cube-mcp/): Documented clients are Claude, ChatGPT, Microsoft 365 Copilot, Cursor and Codex, plus other MCP clients over HTTPS. Admins choose the default deployment and can limit which deployments MCP clients can reach, always intersected with role-based permissions. - [DataHub MCP Server](/servers/datahub-mcp/): The README says the server shows data in databases, lakes, warehouses and BI tools registered in DataHub; no connector list is on this page. - [Euno MCP](/servers/euno-mcp/): The published tool table covers metadata, lineage, EQL, audit, documentation, catalog and benchmark tools; no listed tool executes SQL against a warehouse, and query_to_eql converts a filter without executing it. Euno's FAQ says Euno does not query customer data. - [Hex MCP server](/servers/hex-mcp/): The server works with Claude Desktop, Claude Code, Cursor, ChatGPT, Codex and Glean as well as standard MCP clients. Single-tenant, EU multi-tenant and HIPAA multi-tenant workspaces use their own host instead of app.hex.tech. - [Looker tools in MCP Toolbox for Databases](/servers/looker-mcp-toolbox/): Looker can be deployed in the cloud, on GCP, or on premises, and the same Toolbox covers Google Cloud databases such as AlloyDB, BigQuery, Cloud SQL, Spanner and Firestore. Looker connection tools list databases, schemas, tables and columns of the warehouse connections configured in Looker. Toolbox runs as a local stdio MCP server through npx @toolbox-sdk/server with the --stdio flag. - [Omni MCP Server](/servers/omni-mcp/): The same MCP server can serve embed users, scoped to each user's permissions. Setup pages exist for ChatGPT, Claude Code, Claude Desktop, Codex, Cursor, Microsoft Copilot and VS Code. - [OpenMetadata MCP Server](/servers/openmetadata-mcp/): The reference lists asset types: database, dashboard, pipeline, topic, mlmodel, container, searchIndex, storedProcedure and API services; no named vendor list is on this page. - [Power BI Modeling MCP Server](/servers/powerbi-modeling-mcp/): The local server connects to semantic models in Power BI Desktop, Fabric workspaces and Power BI Project (PBIP) files; --compatibility Full also targets Analysis Services databases. Hosted uses Streamable HTTP and Entra ID; local uses stdio, with Desktop, PBIP files, transactions and traces only on local. - [Secoda MCP](/servers/secoda-mcp/): None found: A list of supported warehouses or BI tools on the Secoda MCP page. - [Select Star MCP Server](/servers/select-star-mcp/): Select Star says its universal metadata model connects more than 25 integrations (founder post on the Snowflake deal); the MCP page names no systems. - [Sigma MCP server](/servers/sigma-mcp/): Documented clients include Claude, ChatGPT, Codex, Cortex Code and Cursor over HTTP transport. The configuration table sets transport to HTTP. - [Tableau MCP](/servers/tableau-mcp/): The server works with Tableau Cloud and Tableau Server of any edition, with some tools gated by SKU. Tools span published data sources, workbooks, views, custom views, projects, Prep flows, Pulse metrics, extract refresh tasks, users and Admin Insights. The server speaks stdio by default and HTTP as an alternative transport. - [ThoughtSpot MCP Server](/servers/thoughtspot-mcp-server/): The tool set includes connectivity and auth checks (ping, check_connectivity) and org handling tools list_orgs and switch_org, so one server can reach multiple ThoughtSpot Orgs. The README lists /mcp, /sse, /token/* and /bearer/* endpoints on the Worker. **What Euno's server does not do.** Euno's server gives an agent governed context about the data; it does not run the query, which is the job of the warehouse's own server. ## Other situations - [A data platform lead whose analysts no longer trust the dashboards](/situations/untrusted-dashboards/) - [An analytics engineering team on dbt and Snowflake or Databricks](/situations/dbt-warehouse/) - [A company putting AI agents on its data this year](/situations/agents-this-year/) - [A governance lead who must show where a regulated field flows](/situations/regulated-field/) ## Sources - github.com/Alation/alation-ai-agent-sdk/blob/main/README.md: https://github.com/Alation/alation-ai-agent-sdk/blob/main/README.md (read 10 Nov 2026) - docs.atlan.com/product/capabilities/atlan-ai/how-tos/remote-mcp-overview: https://docs.atlan.com/product/capabilities/atlan-ai/how-tos/remote-mcp-overview (read 10 Nov 2026) - github.com/collibra/chip/blob/main/README.md: https://github.com/collibra/chip/blob/main/README.md (read 10 Nov 2026) - docs.cube.dev/docs/integrations/mcp-server.md: https://docs.cube.dev/docs/integrations/mcp-server.md (read 10 Nov 2026) - raw.githubusercontent.com/acryldata/mcp-server-datahub/main/README.md: https://raw.githubusercontent.com/acryldata/mcp-server-datahub/main/README.md (read 10 Nov 2026) - docs.euno.ai/ai/mcp-setup.md: https://docs.euno.ai/ai/mcp-setup.md (read 10 Nov 2026) - euno.ai: https://euno.ai/ (read 10 Nov 2026) - learn.hex.tech/docs/api-integrations/mcp-server: https://learn.hex.tech/docs/api-integrations/mcp-server (read 10 Nov 2026) - mcp-toolbox.dev/integrations/looker/source: https://mcp-toolbox.dev/integrations/looker/source/ (read 10 Nov 2026) - mcp-toolbox.dev/llms.txt: https://mcp-toolbox.dev/llms.txt (read 10 Nov 2026) - github.com/googleapis/mcp-toolbox: https://github.com/googleapis/mcp-toolbox (read 10 Nov 2026) - docs.omni.co/llms.txt: https://docs.omni.co/llms.txt (read 10 Nov 2026) - docs.omni.co/sitemap.xml: https://docs.omni.co/sitemap.xml (read 10 Nov 2026) - docs.open-metadata.org/v2.0.x/how-to-guides/mcp/reference.md: https://docs.open-metadata.org/v2.0.x/how-to-guides/mcp/reference.md (read 10 Nov 2026) - raw.githubusercontent.com/microsoft/powerbi-modeling-mcp/main/README.md: https://raw.githubusercontent.com/microsoft/powerbi-modeling-mcp/main/README.md (read 10 Nov 2026) - learn.microsoft.com/en-us/power-bi/developer/mcp/power-bi-authoring-mcp: https://learn.microsoft.com/en-us/power-bi/developer/mcp/power-bi-authoring-mcp (read 10 Nov 2026) - docs.secoda.co/features/ai-assistant/secoda-mcp-server.md: https://docs.secoda.co/features/ai-assistant/secoda-mcp-server.md (read 10 Nov 2026) - selectstar.com/resources/snowflake-acquire-select-star: https://selectstar.com/resources/snowflake-acquire-select-star (read 10 Nov 2026) - help.sigmacomputing.com/docs/use-sigma-mcp-server: https://help.sigmacomputing.com/docs/use-sigma-mcp-server (read 10 Nov 2026) - tableau.github.io/tableau-mcp/docs/intro: https://tableau.github.io/tableau-mcp/docs/intro (read 10 Nov 2026) - tableau.github.io/tableau-mcp/docs/configuration/mcp-config/env-vars: https://tableau.github.io/tableau-mcp/docs/configuration/mcp-config/env-vars (read 10 Nov 2026) - raw.githubusercontent.com/thoughtspot/mcp-server/main/src/servers/tool-definitions.ts: https://raw.githubusercontent.com/thoughtspot/mcp-server/main/src/servers/tool-definitions.ts (read 10 Nov 2026) - raw.githubusercontent.com/thoughtspot/mcp-server/main/README.md: https://raw.githubusercontent.com/thoughtspot/mcp-server/main/README.md (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/situations/bi-sprawl/ --- # Five situations, the checks each one needs, and the facts that answer them SITUATIONS, read 10 Nov 2026 **The short answer.** Five situations an engineer meets when wiring an agent to data: dashboards nobody trusts, dbt on a warehouse, agents on the data this year, a regulated field, and BI sprawl. Each page lists what to check, which of the fourteen fields answers it, and those fields for the servers in scope, read 10 Nov 2026, naming no choice. - [A data platform lead whose analysts no longer trust the dashboards](/situations/untrusted-dashboards/): Check lineage, impact and incident tools across catalogs, observability and BI servers. - [An analytics engineering team on dbt and Snowflake or Databricks](/situations/dbt-warehouse/): Check the dbt, Snowflake and Databricks servers side by side: what runs SQL, what reads models. - [A company putting AI agents on its data this year](/situations/agents-this-year/): Check transport, auth as the user, write tools and tool-definition size across every server. - [A governance lead who must show where a regulated field flows](/situations/regulated-field/): Check lineage depth, classification tools, audit and scoping in catalog and warehouse servers. - [A BI team with thousands of reports and no map of what feeds them](/situations/bi-sprawl/): Check what each BI server exposes about data sources, semantic models and lineage. Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/situations/ --- # The published tool definitions of 24 MCP servers measure from 63 to 4,550 tokens CONTEXT WEIGHT INDEX, read 10 Nov 2026 **The short answer.** The Context Weight Index counts the tokens in each server's published tool names and descriptions, one method for all, read 10 Nov 2026. 24 servers were measured: from 63 tokens (Secoda MCP) to 4,550 (SYNQ MCP (Coalesce Quality)), median 662. 6 publish no measurable tool list. Every tool an MCP server lists is described to the agent in its tool definition: a name, a description and an input schema. A client loads those definitions into the model's context window when it connects, before the agent does any work. The Context Weight Index measures how large each server's published tool definitions are, in tokens, with one method for every server. The table is filed by data tool. Select a column heading to sort by it; the CSV and JSON downloads carry the same rows. | Data tool | Server | Tools measured | With a description | Tokens | Tokens per tool | Source measured | | --- | --- | --- | --- | --- | --- | --- | | Airbyte | [Airbyte MCP (hosted)](/servers/airbyte-mcp/) | 50 | 50 | 870 | 17 | docs page | | Alation | [Alation AI Agent SDK MCP server](/servers/alation-mcp/) | 4 | 3 | 69 | 17 | docs page (parameters published, not counted) | | Amazon Redshift | [Amazon Redshift MCP Server (AWS Labs)](/servers/awslabs-redshift-mcp-server/) | 8 | 8 | 238 | 30 | README (parameters published, not counted) | | Atlan | [Atlan MCP Server](/servers/atlan-mcp/) | 42 | 42 | 1770 | 42 | docs page (parameters published, not counted) | | BigQuery | [BigQuery remote MCP server](/servers/bigquery-remote-mcp/) | 9 | 9 | 1487 | 165 | docs page | | BigQuery | [MCP Toolbox for Databases (BigQuery tools)](/servers/mcp-toolbox-bigquery/) | 9 | 9 | 156 | 17 | docs page | | Collibra | [Collibra MCP Server (chip)](/servers/collibra-mcp/) | 41 | 41 | 2914 | 71 | README | | Cube | [Cube MCP server](/servers/cube-mcp/) | 30 | 30 | 1093 | 36 | docs page | | Databricks | [Databricks managed MCP servers](/servers/databricks-managed-mcp/) | 4 | 4 | 94 | 24 | docs page | | DataHub | [DataHub MCP Server](/servers/datahub-mcp/) | 22 | 22 | 567 | 26 | README | | dbt | [dbt MCP Server](/servers/dbt-mcp/) | 63 | 63 | 1061 | 17 | README | | Euno | [Euno MCP](/servers/euno-mcp/) | 56 | 42 | 760 | 14 | docs page | | Fivetran | [Fivetran MCP Server](/servers/fivetran-mcp/) | 53 | 2 | 299 | 6 | README | | Hex | [Hex MCP server](/servers/hex-mcp/) | 18 | 18 | 252 | 14 | docs page | | Looker | [Looker tools in MCP Toolbox for Databases](/servers/looker-mcp-toolbox/) | 55 | 55 | 1475 | 27 | docs index | | Omni | [Omni MCP Server](/servers/omni-mcp/) | 27 | 27 | 466 | 17 | docs page | | OpenMetadata | [OpenMetadata MCP Server](/servers/openmetadata-mcp/) | 15 | 15 | 1249 | 83 | docs page (parameters published, not counted) | | Power BI | [Power BI Modeling MCP Server](/servers/powerbi-modeling-mcp/) | 21 | 21 | 387 | 18 | README | | Secoda | [Secoda MCP](/servers/secoda-mcp/) | 6 | 6 | 63 | 11 | source code (parameters published, not counted) | | Select Star | [Select Star MCP Server](/servers/select-star-mcp/) | 7 | 7 | 98 | 14 | docs page | | Sifflet | [Sifflet MCP Server](/servers/sifflet-mcp/) | 10 | 10 | 662 | 66 | source code (parameters published, not counted) | | Snowflake | [mcp-snowflake-server (community)](/servers/isaacwasserman-mcp-snowflake-server/) | 8 | 8 | 103 | 13 | README (parameters published, not counted) | | SYNQ | [SYNQ MCP (Coalesce Quality)](/servers/synq-mcp/) | 134 | 134 | 4550 | 34 | docs page | | ThoughtSpot | [ThoughtSpot MCP Server](/servers/thoughtspot-mcp-server/) | 20 | 20 | at least 2149 | 107 | source code (parameters published, not counted) | ## Servers that could not be measured - [Databricks Labs MCP servers](/servers/databrickslabs-mcp/): README says the tool list is inferred at startup from the Unity Catalog functions, vector search indexes and Genie spaces the user configures, so no fixed per-tool list or descriptions are published. - [Monte Carlo MCP Server](/servers/monte-carlo-mcp/): no published tool-definition text was read on 10 Nov 2026 - [Sigma MCP server](/servers/sigma-mcp/): The docs page describes features (search, explore, analyze, build workbooks, screenshots, agents) in prose with example prompts but publishes no tool names or per-tool list. - [Snowflake Labs MCP Server](/servers/snowflake-labs-mcp/): The Snowflake Labs server, marked deprecated in its README, which describes config-driven tool groups (Cortex Search, Analyst, Agent, object management, SQL execution, semantic views) but publishes no per-tool names or descriptions; facts.json lists these six groups. - [Snowflake-managed MCP server](/servers/snowflake-managed-mcp/): The docs page defines user-configured tool types (CORTEX_AGENT_RUN, CODING_AGENT, CORTEX_SEARCH_SERVICE_QUERY, CORTEX_ANALYST_MESSAGE, SYSTEM_EXECUTE_SQL, GENERIC) used to build tools in a CREATE MCP SERVER specification; it publishes no fixed list of named tools, so facts.json lists tool types not tools. - [Tableau MCP](/servers/tableau-mcp/): the source read on 10 Nov 2026 holds the definitions of 2 of the roughly 40 Tableau MCP tools (get-datasource-metadata and query-datasource), and one of those descriptions is imported from a file that was not read, so a measurement would cover a small part of the server. ## Partial measurements - [ThoughtSpot MCP Server](/servers/thoughtspot-mcp-server/): Lower bound: eight of the 20 descriptions are built from several string literals or constants in the source, and only the first literal of each was measured. ## What the number is, and what it is not The index measures the tool names and descriptions each publisher prints in its own source, read on 10 Nov 2026. The same counting rule is applied to each server's own published text; the kind of text differs (a docs page, a README or source code), and the table shows which kind was measured for each server. It does not include input schemas, because most sources do not publish them; where they are published, the table says so. A docs page often prints a one-line summary where the server itself sends a longer description, so the number can be lower than what a client loads at connection time. A server that lists more tools than an agent needs can usually be narrowed with the allow-lists, toolsets or modes its own documentation describes; the entries record those switches where they are documented. Tokens are counted with the o200k_base encoding. Other models' tokenizers give different absolute counts; the relative sizes stay close. ## Cite this Data MCP Directory, The Context Weight Index, datamcpservers.com/context-weight/, edition of 10 Nov 2026, CC BY 4.0. ## Sources - Airbyte MCP (hosted): docs.airbyte.com/platform/airbyte-mcp/tools.md: https://docs.airbyte.com/platform/airbyte-mcp/tools.md (read 10 Nov 2026) - Alation AI Agent SDK MCP server: github.com/Alation/alation-ai-agent-sdk/tree/main/guides: https://github.com/Alation/alation-ai-agent-sdk/tree/main/guides (read 10 Nov 2026) - Amazon Redshift MCP Server (AWS Labs): raw.githubusercontent.com/awslabs/mcp/main/src/redshift-mcp-server/README.md: https://raw.githubusercontent.com/awslabs/mcp/main/src/redshift-mcp-server/README.md (read 10 Nov 2026) - Atlan MCP Server: docs.atlan.com/product/capabilities/atlan-ai/references/mcp-tools: https://docs.atlan.com/product/capabilities/atlan-ai/references/mcp-tools (read 10 Nov 2026) - BigQuery remote MCP server: docs.cloud.google.com/bigquery/docs/reference/mcp: https://docs.cloud.google.com/bigquery/docs/reference/mcp (read 10 Nov 2026) - MCP Toolbox for Databases (BigQuery tools): raw.githubusercontent.com/googleapis/mcp-toolbox/main/docs/en/integrations/bigquery/prebuilt-configs/bigquery.md: https://raw.githubusercontent.com/googleapis/mcp-toolbox/main/docs/en/integrations/bigquery/prebuilt-configs/bigquery.md (read 10 Nov 2026) - Collibra MCP Server (chip): github.com/collibra/chip/blob/main/README.md: https://github.com/collibra/chip/blob/main/README.md (read 10 Nov 2026) - Cube MCP server: docs.cube.dev/docs/integrations/mcp-server.md: https://docs.cube.dev/docs/integrations/mcp-server.md (read 10 Nov 2026) - Databricks managed MCP servers: docs.databricks.com/aws/en/generative-ai/mcp/managed-mcp: https://docs.databricks.com/aws/en/generative-ai/mcp/managed-mcp (read 10 Nov 2026) - DataHub MCP Server: github.com/acryldata/mcp-server-datahub/blob/main/README.md: https://github.com/acryldata/mcp-server-datahub/blob/main/README.md (read 10 Nov 2026) - dbt MCP Server: raw.githubusercontent.com/dbt-labs/dbt-mcp/main/README.md: https://raw.githubusercontent.com/dbt-labs/dbt-mcp/main/README.md (read 10 Nov 2026) - Euno MCP: docs.euno.ai/ai/mcp-setup.md: https://docs.euno.ai/ai/mcp-setup.md (read 10 Nov 2026) - Fivetran MCP Server: raw.githubusercontent.com/fivetran/fivetran-mcp/main/README.md: https://raw.githubusercontent.com/fivetran/fivetran-mcp/main/README.md (read 10 Nov 2026) - Hex MCP server: learn.hex.tech/docs/api-integrations/mcp-server: https://learn.hex.tech/docs/api-integrations/mcp-server (read 10 Nov 2026) - Looker tools in MCP Toolbox for Databases: mcp-toolbox.dev/llms.txt: https://mcp-toolbox.dev/llms.txt (read 10 Nov 2026) - Omni MCP Server: docs.omni.co/ai/mcp/tools.md: https://docs.omni.co/ai/mcp/tools.md (read 10 Nov 2026) - OpenMetadata MCP Server: docs.open-metadata.org/v2.0.x/how-to-guides/mcp/reference.md: https://docs.open-metadata.org/v2.0.x/how-to-guides/mcp/reference.md (read 10 Nov 2026) - Power BI Modeling MCP Server: raw.githubusercontent.com/microsoft/powerbi-modeling-mcp/main/README.md: https://raw.githubusercontent.com/microsoft/powerbi-modeling-mcp/main/README.md (read 10 Nov 2026) - Secoda MCP: github.com/secoda/secoda-mcp/blob/main/server.py: https://github.com/secoda/secoda-mcp/blob/main/server.py (read 10 Nov 2026) - Select Star MCP Server: docs.selectstar.com/features/mcp-server: https://docs.selectstar.com/features/mcp-server (read 10 Nov 2026) - Sifflet MCP Server: raw.githubusercontent.com/siffletdata/sifflet-mcp/1f390ca06d7919214bfe067991539502fdb4b9fc/src/sifflet_mcp/server.py: https://raw.githubusercontent.com/siffletdata/sifflet-mcp/1f390ca06d7919214bfe067991539502fdb4b9fc/src/sifflet_mcp/server.py (read 10 Nov 2026) - mcp-snowflake-server (community): raw.githubusercontent.com/isaacwasserman/mcp-snowflake-server/main/README.md: https://raw.githubusercontent.com/isaacwasserman/mcp-snowflake-server/main/README.md (read 10 Nov 2026) - SYNQ MCP (Coalesce Quality): docs.synq.io/scout/mcp.md: https://docs.synq.io/scout/mcp.md (read 10 Nov 2026) - ThoughtSpot MCP Server: raw.githubusercontent.com/thoughtspot/mcp-server/main/src/servers/tool-definitions.ts: https://raw.githubusercontent.com/thoughtspot/mcp-server/main/src/servers/tool-definitions.ts (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/context-weight/ --- # One rule for every server: count the published name and description of each tool, nothing else INDEX METHODOLOGY, read 10 Nov 2026 **The short answer.** For every server, the index takes each tool the server's own source lists, writes it as its name and its published description, joins the lines and counts tokens with the o200k_base encoding. A script checks every description word for word against the source text read on 10 Nov 2026, and fails the build on any mismatch. The rule is the same for every server. For each tool the server's own source lists, the text measured is the tool's name, a colon and a space, and the description the source gives for that tool, word for word. A tool listed with no description counts as its name alone. The lines are joined with newlines and the tokens counted with the o200k_base encoding (js-tiktoken 1.0.21). The source is the one recorded on each entry: the server's README, its source code, or its vendor documentation, with the version or commit read. Deprecated tools still listed in the source are counted, because a client still loads them. Before counting, a script checks that every description appears verbatim in the saved source text, after collapsing whitespace and removing markdown backticks, asterisks and table pipes. Any mismatch fails the run, so no description is paraphrased or shortened. A server is marked not measured when its source publishes no per-tool list (tools generated at startup from the user's own functions, configuration-driven tool groups, prose without tool names) or when the source read holds only a small part of the list. A measurement built from the first of several joined string literals is marked as a lower bound. ## The source measured for each server | Server | Source | Version read | Note | | --- | --- | --- | --- | | [Airbyte MCP (hosted)](/servers/airbyte-mcp/) | [docs.airbyte.com/platform/airbyte-mcp/tools.md](https://docs.airbyte.com/platform/airbyte-mcp/tools.md) | hosted service (no versioned release; docs page) | Docs page lists 50 tools (same count as facts.json) with one-line descriptions in tables grouped by area; parameters are published only for execute_external_search_query and request_search_index_backfill, not for the other tools. | | [Alation AI Agent SDK MCP server](/servers/alation-mcp/) | [github.com/Alation/alation-ai-agent-sdk/tree/main/guides](https://github.com/Alation/alation-ai-agent-sdk/tree/main/guides) | alation-ai-agent-mcp 1.0.0 (tool guides lineage, catalog_context_search_agent, get_data_products, data_quality_tool) | Source is four SDK guide pages (three tools and one agent) with input parameters for lineage, catalog_context_search_agent and get_data_products; data_quality_tool has no one-line description (its guide opens with Background), and the list is shorter than the 12 names in facts.json because only these four guides were published. | | [Amazon Redshift MCP Server (AWS Labs)](/servers/awslabs-redshift-mcp-server/) | [raw.githubusercontent.com/awslabs/mcp/main/src/redshift-mcp-server/README.md](https://raw.githubusercontent.com/awslabs/mcp/main/src/redshift-mcp-server/README.md) | PyPI 0.1.3 | README Tools section gives, for each tool, a description, a typed Python signature, a Parameters list and Returns. | | [Atlan MCP Server](/servers/atlan-mcp/) | [docs.atlan.com/product/capabilities/atlan-ai/references/mcp-tools](https://docs.atlan.com/product/capabilities/atlan-ai/references/mcp-tools) | hosted Atlan MCP, page last modified 27 Oct 2026 | Hosted docs page gives 41 tool names with one-line descriptions and a Read/Write/Admin label, no parameters; the file also includes the deprecated local server code for traverse_lineage_tool (with typed parameters and a docstring), listed here as a 42nd entry; the entry names the 41 hosted tools, and the 42nd measured here is the deprecated local traverse_lineage_tool, still listed in the source. | | [BigQuery remote MCP server](/servers/bigquery-remote-mcp/) | [docs.cloud.google.com/bigquery/docs/reference/mcp](https://docs.cloud.google.com/bigquery/docs/reference/mcp) | docs last updated 3 Nov 2026 | Google Cloud docs page lists tool names with long descriptions (including example queries) but no parameter table; parameter names appear only inside description text. | | [MCP Toolbox for Databases (BigQuery tools)](/servers/mcp-toolbox-bigquery/) | [raw.githubusercontent.com/googleapis/mcp-toolbox/main/docs/en/integrations/bigquery/prebuilt-configs/bigquery.md](https://raw.githubusercontent.com/googleapis/mcp-toolbox/main/docs/en/integrations/bigquery/prebuilt-configs/bigquery.md) | v1.16.0 | Prebuilt bigquery config lists nine tools with one-line descriptions; parameters are published only for three tools in the per-tool pages, not for all nine, so params_in_source is false. | | [Collibra MCP Server (chip)](/servers/collibra-mcp/) | [github.com/collibra/chip/blob/main/README.md](https://github.com/collibra/chip/blob/main/README.md) | v0.0.67 (README tool list) | README lists tool names with one-line descriptions under Read Tools and Write Tools (some flagged Requires scopes or Gated on the data-quality capability), without parameter schemas, though some descriptions mention parameter names. | | [Cube MCP server](/servers/cube-mcp/) | [docs.cube.dev/docs/integrations/mcp-server.md](https://docs.cube.dev/docs/integrations/mcp-server.md) | hosted service (no versioned release; docs page, lines 517-740 'Available actions') | Docs page states 30 tools and lists each with a one-line description and an access label (read-only, write or destructive) in tables grouped by area; parameter names appear only inside description and prose text. | | [Databricks Labs MCP servers](/servers/databrickslabs-mcp/) | [raw.githubusercontent.com/databrickslabs/mcp/master/README.md](https://raw.githubusercontent.com/databrickslabs/mcp/master/README.md) | pushed 26 Oct 2026 | README says the tool list is inferred at startup from the Unity Catalog functions, vector search indexes and Genie spaces the user configures, so no fixed per-tool list or descriptions are published. | | [Databricks managed MCP servers](/servers/databricks-managed-mcp/) | [docs.databricks.com/aws/en/generative-ai/mcp/managed-mcp](https://docs.databricks.com/aws/en/generative-ai/mcp/managed-mcp) | docs page last updated Nov 9, 2026 | Docs page lists four system.ai MCP endpoints with one-line descriptions and no tool-level definitions or parameters; seven external-service MCPs (Slack, GitHub, Atlassian, Google Drive, Google Calendar, Gmail, Microsoft 365) are named with no description and are not included. | | [DataHub MCP Server](/servers/datahub-mcp/) | [github.com/acryldata/mcp-server-datahub/blob/main/README.md](https://github.com/acryldata/mcp-server-datahub/blob/main/README.md) | 0.8.0 (tool list from README; registration code from src/mcp_server_datahub/mcp_server.py) | README lists tool names with one-line descriptions (paired tools such as add_tags / remove_tags share one description, copied to each); the appended registration code names tools but adds no parameters or per-tool descriptions beyond a templated docstring for search, which is not included. | | [dbt MCP Server](/servers/dbt-mcp/) | [raw.githubusercontent.com/dbt-labs/dbt-mcp/main/README.md](https://raw.githubusercontent.com/dbt-labs/dbt-mcp/main/README.md) | PyPI 2.10.1 | README lists tool names with one-line descriptions grouped by area; deprecated tools (get_exposure_details, get_macro_details, get_model_children, get_model_details, get_model_parents, get_seed_details, get_semantic_model_details, get_snapshot_details, get_source_details, get_test_details) are included with their description being only the deprecation notice, and search is marked Alpha. | | [Euno MCP](/servers/euno-mcp/) | [docs.euno.ai/ai/mcp-setup.md](https://docs.euno.ai/ai/mcp-setup.md?utm_source=datamcpservers.com&utm_medium=referral&utm_campaign=method) | (hosted service, no version; table of tool names and descriptions only, no parameter schemas) | Docs page lists 42 tools in a table with one-line descriptions and an availability column, plus 14 catalog tools (glossary, metadata tag and workflow create/fetch/find/update/delete) named in a bullet list with no description, which have an empty description here; no parameter schemas are published. | | [Fivetran MCP Server](/servers/fivetran-mcp/) | [raw.githubusercontent.com/fivetran/fivetran-mcp/main/README.md](https://raw.githubusercontent.com/fivetran/fivetran-mcp/main/README.md) | v0.5.1 | README describes list_endpoints and get_schema in prose (descriptions start lowercase because they follow the tool name), and a generated table names 51 resource_action execution tools with endpoint counts but no per-tool description (empty here), so the count differs from the 3 entries in facts.json (which lists the family as one item); tools shown are filtered at runtime by FIVETRAN_SCOPE and DISALLOWED_ACTIONS. | | [Hex MCP server](/servers/hex-mcp/) | [learn.hex.tech/docs/api-integrations/mcp-server](https://learn.hex.tech/docs/api-integrations/mcp-server) | hosted service (no versioned release; docs page tool list) | Docs page lists 18 tools in prose in two groups (knowledge tools and project editing tools); slash-joined tools (for example create_cell / update_cell / delete_cell) share one description, copied to each, and no parameters are published. | | [Looker tools in MCP Toolbox for Databases](/servers/looker-mcp-toolbox/) | [mcp-toolbox.dev/llms.txt](https://mcp-toolbox.dev/llms.txt) | v1.16.0 (fa59758028ec) | The llms.txt index publishes one-line descriptions for 55 Looker tools (names taken from the page URLs); the count matches facts.json. | | [Monte Carlo MCP Server](/servers/monte-carlo-mcp/) | none read | none | no published tool-definition text was read on 10 Nov 2026 | | [Omni MCP Server](/servers/omni-mcp/) | [docs.omni.co/ai/mcp/tools.md](https://docs.omni.co/ai/mcp/tools.md) | hosted service (no versioned release; docs page) | Docs page lists 26 tools with an accordion one-line summary each, plus searchOmniDocs and queryUsageAnalytics described in their own sections (the two import tools have no summary line so their first paragraph is used); facts.json lists 26 (without searchOmniDocs) so the raw file was followed; some optional arguments are mentioned in prose but no input schemas are published. | | [OpenMetadata MCP Server](/servers/openmetadata-mcp/) | [docs.open-metadata.org/v2.0.x/how-to-guides/mcp/reference.md](https://docs.open-metadata.org/v2.0.x/how-to-guides/mcp/reference.md) | OpenMetadata 2.0.x docs | Docs reference page gives a Description, Read/Write type, parameter tables and JSON examples for each of 15 tools; matches facts.json. | | [Power BI Modeling MCP Server](/servers/powerbi-modeling-mcp/) | [raw.githubusercontent.com/microsoft/powerbi-modeling-mcp/main/README.md](https://raw.githubusercontent.com/microsoft/powerbi-modeling-mcp/main/README.md) | v1.0.1 (2b25defb9751) | README Available tools table with tool names and one-line descriptions; each tool is an operation group covering several operations; no parameters published. | | [Secoda MCP](/servers/secoda-mcp/) | [github.com/secoda/secoda-mcp/blob/main/server.py](https://github.com/secoda/secoda-mcp/blob/main/server.py) | commit of 26 Jan 2026 | server.py source defines 6 tools with Python docstring descriptions and typed parameters; facts.json lists 8 (search, get_secoda_docs and chart are not in this source file, and search_documentation is not in facts.json), so the raw file was followed. | | [Select Star MCP Server](/servers/select-star-mcp/) | [docs.selectstar.com/features/mcp-server](https://docs.selectstar.com/features/mcp-server) | docs page read 10 Nov 2026 via web.archive.org snapshot 20260614045618 | Docs page lists seven tool names, each with a one-line description; no parameters published. | | [Sifflet MCP Server](/servers/sifflet-mcp/) | [raw.githubusercontent.com/siffletdata/sifflet-mcp/1f390ca06d7919214bfe067991539502fdb4b9fc/src/sifflet_mcp/server.py](https://raw.githubusercontent.com/siffletdata/sifflet-mcp/1f390ca06d7919214bfe067991539502fdb4b9fc/src/sifflet_mcp/server.py) | 1f390ca (PyPI 0.1.7, 3 Aug 2026) | server.py source defines ten tools with multi-line description strings and typed Python parameters (some descriptions also explain parameter values); matches facts.json. | | [Sigma MCP server](/servers/sigma-mcp/) | [help.sigmacomputing.com/docs/use-sigma-mcp-server.md](https://help.sigmacomputing.com/docs/use-sigma-mcp-server.md) | hosted service (no versioned release); features text only, tool names not published on the page | The docs page describes features (search, explore, analyze, build workbooks, screenshots, agents) in prose with example prompts but publishes no tool names or per-tool list. | | [mcp-snowflake-server (community)](/servers/isaacwasserman-mcp-snowflake-server/) | [raw.githubusercontent.com/isaacwasserman/mcp-snowflake-server/main/README.md](https://raw.githubusercontent.com/isaacwasserman/mcp-snowflake-server/main/README.md) | PyPI 0.4.0 | README lists 8 tools with a description, typed Input list and Returns for each; write_query and create_table are enabled only with --allow-write; the two MCP resources (memo://insights and context://table/{table_name}) are not tools and are not included. | | [Snowflake Labs MCP Server](/servers/snowflake-labs-mcp/) | [raw.githubusercontent.com/Snowflake-Labs/mcp/main/README.md](https://raw.githubusercontent.com/Snowflake-Labs/mcp/main/README.md) | PyPI 1.4.2 (deprecated; config-driven tools) | The Snowflake Labs server, marked deprecated in its README, which describes config-driven tool groups (Cortex Search, Analyst, Agent, object management, SQL execution, semantic views) but publishes no per-tool names or descriptions; facts.json lists these six groups. | | [Snowflake-managed MCP server](/servers/snowflake-managed-mcp/) | [docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-mcp](https://docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-mcp) | docs page (MCP revision 25 Nov 2025) | The docs page defines user-configured tool types (CORTEX_AGENT_RUN, CODING_AGENT, CORTEX_SEARCH_SERVICE_QUERY, CORTEX_ANALYST_MESSAGE, SYSTEM_EXECUTE_SQL, GENERIC) used to build tools in a CREATE MCP SERVER specification; it publishes no fixed list of named tools, so facts.json lists tool types not tools. | | [SYNQ MCP (Coalesce Quality)](/servers/synq-mcp/) | [docs.synq.io/scout/mcp.md](https://docs.synq.io/scout/mcp.md) | Coalesce Quality (SYNQ) MCP docs | Docs page lists 134 tools in grouped tables with one-line descriptions (no parameter schemas, though some descriptions mention argument names). | | [Tableau MCP](/servers/tableau-mcp/) | [raw.githubusercontent.com/tableau/tableau-mcp/main/src/tools/web/{getDatasourceMetadata/getDatasourceMetadata.ts,queryDatasource/queryDatasource.ts}](https://raw.githubusercontent.com/tableau/tableau-mcp/main/src/tools/web/{getDatasourceMetadata/getDatasourceMetadata.ts,queryDatasource/queryDatasource.ts}) | 8d672c9ab574 (npm 6.0.4); two of the tool source files only | Not measured: the source read on 10 Nov 2026 holds the definitions of 2 of the roughly 40 Tableau MCP tools (get-datasource-metadata and query-datasource), and one of those descriptions is imported from a file that was not read, so a measurement would cover a small part of the server. | | [ThoughtSpot MCP Server](/servers/thoughtspot-mcp-server/) | [raw.githubusercontent.com/thoughtspot/mcp-server/main/src/servers/tool-definitions.ts](https://raw.githubusercontent.com/thoughtspot/mcp-server/main/src/servers/tool-definitions.ts) | 820bf16f6aa3 (package.json 0.5.0) | tool-definitions.ts source lists 20 tools (V1 to V5 definition sets) with description strings and zod input/output schemas; matches facts.json. Where a description is built from several string literals (search_objects, get_data, the four model tools) or an interpolated constant (modify_dashboard, get_dashboard_status), only the first literal is given so the text stays verbatim; backslash escapes before backticks are kept as written; create_dashboard uses the latest (V5 designer) definition, create_analysis_session the V2 definition (a SpotterX variant adds one sentence), and get_session_updates the shared definition (a raw-updates variant appends text). | Filed by data tool, read 10 Nov 2026. ## The script The script is euno-sites/tools/cwi.mjs in the run's repository; it reads the extracts in euno-sites/data/cwi/extract/, checks each description against research/raw/schemas/, and writes data/cwi/index.json, which the site builds from. The extracts list every tool name and description measured, so anyone can recount. ## Cite this Data MCP Directory, The Context Weight Index, datamcpservers.com/context-weight/, edition of 10 Nov 2026, CC BY 4.0. ## Sources - Context Weight Index, CSV: https://datamcpservers.com/datasets/context-weight-index.csv (read 10 Nov 2026) - Context Weight Index, JSON: https://datamcpservers.com/datasets/context-weight-index.json (read 10 Nov 2026) - tiktoken encodings (o200k_base): https://github.com/dqbd/tiktoken (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/context-weight/methodology/ --- # An MCP server for data exposes up to five kinds of tools, and each server's tool list shows which kinds it actually ships GUIDE, read 10 Nov 2026 **The short answer.** Agents meet five kinds of tools on data servers: query, metadata, lineage, semantic layer and write actions. The dbt MCP Server lists tools in all five, BigQuery's remote server lists 9 tools with none for lineage, and the AWS Labs Redshift server lists 8. The tool list read on 10 Nov 2026 shows which kinds each server ships. ## Query tools run SQL or a modeled query and return rows A query tool sends a statement or a structured request to an engine and returns rows. It is the kind of tool most engineers expect first, and the servers split on what the statement is. Google's BigQuery reference lists two SQL tools: execute_sql_readonly, which Google says is restricted to SELECT statements, and execute_sql, which can run INSERT, UPDATE, DELETE and CREATE statements. The [Amazon Redshift MCP Server from AWS Labs](/servers/awslabs-redshift-mcp-server/) lists one, execute_query. Some catalog servers also carry a query tool. Per Atlan's tools page, query_assets runs read-only SQL against a connected warehouse (Snowflake, Redshift, BigQuery or Databricks) and returns up to 100 rows. Secoda's docs describe run_sql as executing SQL against connected warehouses. Databricks documents a dbsql MCP that runs SQL under the caller's Unity Catalog and warehouse permissions; per Databricks, dbsql sits behind the Unity Gateway beta. A modeled query is the other form. Per Cube's docs, runQuery runs Cube SQL against views whose names and members come from searchDataModel, so the agent writes against the semantic model and not against raw tables. ## Metadata tools describe schemas, assets and owners A metadata tool answers what exists and what it means without returning business rows. The Redshift server lists list_clusters, list_databases, list_schemas, list_tables and list_columns, which walk from cluster to column. BigQuery's remote server lists list_dataset_ids, get_dataset_info, list_table_ids and get_table_info. The depth differs. Per the community mcp-snowflake-server README, describe_table returns column names, types, nullability, defaults and comments, so table and column comments reach the agent. Per Tableau's docs, get-datasource-metadata returns table relationships, field descriptions, roles and types, calculation strings and parameters for a data source. [The DataHub MCP Server](/servers/datahub-mcp/) lists list_schema_fields and get_entities, and its README says agents can discover domains, owners, tags and glossary terms. Per Collibra's README, discover_business_glossary and get_column_semantics reach business terms connected to a column. For these catalog tools the content comes from what the catalog holds, so the README does not list the warehouses behind it (none found on 10 Nov 2026 in the Collibra README). Pipeline and observability tools add a state dimension to metadata. Sifflet's search_asset filters by asset type TABLE_AND_VIEW, PIPELINE, DASHBOARD or ML_MODEL. Airbyte's describe_connection returns a connection's streams, schedule and status, and discover_source_catalog is a catalog tool at stream level. Per Monte Carlo's docs, the extended toolset adds get_use_cases and get_use_case_tables for critical-table groupings. ## Lineage tools trace what feeds an asset and what depends on it Lineage tools return a graph instead of a row set. Per the dbt Labs README, get_lineage returns a bounded graph with type, depth and direction filters, and column-level lineage comes from fusion.get_column_lineage on dbt Platform and get_column_lineage locally. Per DataHub's README, get_lineage_paths_between returns exact paths between two assets or columns, including intermediate transformations and SQL query information. Per Collibra's README, get_lineage_transformation returns the details and logic of a specific transformation. Sifflet's server source lists get_downstream_assets_of_asset, which returns downstream assets by URN. Select Star's page lists traverse_lineage, and [the Select Star MCP Server](/servers/select-star-mcp/) page says nothing on column-level depth (none found on 10 Nov 2026). Per Euno's docs, get_upstream_lineage and resource_impact_analysis work on one resource URI. Warehouse servers tend to publish none. The Redshift server (8 tools), the BigQuery remote server (9 tools), the Snowflake-managed server and the Databricks-provided MCPs list no lineage tool; each page was read on 10 Nov 2026 and the finding is recorded as none found. ## Semantic layer tools return governed definitions instead of raw tables A semantic layer tool hands the agent metrics, dimensions and business definitions, so a question like revenue by region resolves to a defined measure. Per the dbt Labs README, the Semantic Layer tools list metrics, dimensions, entities and saved queries, and query_metrics runs a metric query. Per Cube's docs, searchDataModel searches views and their measures and dimensions by similarity. Snowflake's docs say the Cortex Analyst tools on the Snowflake-managed server use semantic views, which hold business definitions, and that the server accepts semantic views only, not semantic models. Databricks says Genie One resolves business terms, metric definitions and table relationships through Genie Ontology, which it describes as a governed semantic layer. Per Tableau's docs, the server serves Pulse metric definitions so metrics are answered from governed definitions. Per Omni's docs, queries run against modeled topics. The vendors describe these layers in their own words, and the layers are not interchangeable: a dbt metric, a Cube view and a Snowflake semantic view are defined in different places and queried through different tools. ## Write actions change data, models or metadata Write tools are the fifth kind, and they sit inside otherwise read-oriented servers. The dbt Labs README warns that its CLI tools (build, run, test) can modify data models and warehouse objects. Collibra's edit_asset sets attributes such as Definition and assigns Steward and Owner roles, per its README. DataHub's optional tools add tags, terms and owners. Hex lists create_cell, update_cell and delete_cell, and Airbyte lists deploy_connector and permanently_delete_connector. Which servers ship writes, and how each documents confirmation, is covered in the guide on [read versus write](/guides/). This guide only records that the kind exists and that a tool list shows it by name. Write tools are not limited to warehouse data. Per Euno's docs, its catalog tools create, update and delete glossary terms, metadata tags and workflows. ThoughtSpot's tool definitions include createLiveboard, create_dashboard and modify_dashboard, which create or change content inside the BI tool, and Cube's tools create reports and workbooks. ## A tool list shows which kinds a server ships The table puts a sample of servers side by side by the kinds named in their own tool lists or docs. A dash means none found on 10 Nov 2026 in the recorded facts. The table is a reading aid and not a measure of how many tools a server has; the [Context Weight Index](/context-weight/) covers tool-definition size separately. Cells marked none found mean the tool list or docs page read on 10 Nov 2026 named no tool of that kind. They do not say the capability is unavailable through other means, such as the engine's own features. | Server | Query | Metadata | Lineage | Semantic layer | Writes | | --- | --- | --- | --- | --- | --- | | Amazon Redshift MCP Server (AWS Labs) | execute_query | list_clusters to list_columns | none found | none found | read-write mode, opt-in | | BigQuery remote MCP server (Google Cloud) | execute_sql_readonly, execute_sql | get_table_info, get_dataset_info | none found | none found | execute_sql | | Collibra MCP Server | none found | discover_data_assets, get_column_semantics | get_lineage_transformation | get_table_semantics | edit_asset, create_asset | | Cube MCP server | runQuery | searchDataModel | none found | views, measures, dimensions | writeDataModelFile, createReport | | DataHub MCP Server | none found | search, list_schema_fields | get_lineage | none found | add_tags, add_owners (off by default) | | dbt MCP Server | execute_sql, text_to_sql | get_all_models | get_lineage, get_column_lineage | list_metrics, query_metrics | build, run, trigger_job_run | ## Sources - BigQuery MCP tools reference: https://docs.cloud.google.com/bigquery/docs/reference/mcp (read 10 Nov 2026) - AWS Labs Redshift MCP README: https://raw.githubusercontent.com/awslabs/mcp/main/src/redshift-mcp-server/README.md (read 10 Nov 2026) - Atlan MCP tools page: https://docs.atlan.com/product/capabilities/atlan-ai/references/mcp-tools (read 10 Nov 2026) - Secoda MCP server docs: https://docs.secoda.co/features/ai-assistant/secoda-mcp-server.md (read 10 Nov 2026) - Databricks managed MCP page: https://docs.databricks.com/aws/en/generative-ai/mcp/managed-mcp (read 10 Nov 2026) - Cube MCP server docs: https://docs.cube.dev/docs/integrations/mcp-server.md (read 10 Nov 2026) - Community mcp-snowflake-server README: https://raw.githubusercontent.com/isaacwasserman/mcp-snowflake-server/main/README.md (read 10 Nov 2026) - Tableau MCP docs intro: https://tableau.github.io/tableau-mcp/docs/intro (read 10 Nov 2026) - DataHub MCP README: https://raw.githubusercontent.com/acryldata/mcp-server-datahub/main/README.md (read 10 Nov 2026) - Collibra MCP README: https://github.com/collibra/chip/blob/main/README.md (read 10 Nov 2026) - dbt MCP README: https://raw.githubusercontent.com/dbt-labs/dbt-mcp/main/README.md (read 10 Nov 2026) - Sifflet MCP server source: https://github.com/siffletdata/sifflet-mcp/blob/1f390ca06d7919214bfe067991539502fdb4b9fc/src/sifflet_mcp/server.py (read 10 Nov 2026) - Select Star MCP docs: https://docs.selectstar.com/features/mcp-server (read 10 Nov 2026) - Euno MCP setup docs: https://docs.euno.ai/ai/mcp-setup.md (read 10 Nov 2026) - Snowflake-managed MCP docs: https://docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-mcp (read 10 Nov 2026) - Omni MCP tools: https://docs.omni.co/ai/mcp/tools.md (read 10 Nov 2026) - Hex MCP server docs: https://learn.hex.tech/docs/api-integrations/mcp-server (read 10 Nov 2026) - Airbyte MCP tools: https://docs.airbyte.com/platform/airbyte-mcp/tools.md (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/guides/what-an-mcp-server-for-data-should-expose/ --- # Servers differ on whether writes ship at all, whether they are off by default, and what stops a write once an agent calls it GUIDE, read 10 Nov 2026 **The short answer.** Redshift prompts for confirmation on each write, DataHub keeps mutation tools off by default, Fivetran defaults its scope to read, and Select Star lists 7 tools that only search, get or traverse. Cube annotates nine tools as destructive, while Snowflake's docs call annotations client hints and name RBAC as the control. ## Some servers ship read-only by default and need an opt-in for writes The [Amazon Redshift MCP Server](/servers/awslabs-redshift-mcp-server/) is read-only by default per the AWS Labs README, and read-write mode must be selected with ACCESS_MODE. The README adds that the database user's privileges are the security boundary, not the server. Per the README of [the DataHub MCP Server](/servers/datahub-mcp/), mutation tools are disabled by default and need TOOLS_IS_MUTATION_ENABLED=true. The community mcp-snowflake-server disables write operations by default and enables them only with --allow-write, per its README. Collibra's README says its data quality tools (18 tools, several that write or delete) are off by default behind a --data-quality flag. Per the README of [the Fivetran MCP Server](/servers/fivetran-mcp/), FIVETRAN_SCOPE defaults to read and can be raised to read/write or read/write/delete, and a DISALLOWED_ACTIONS list denies resource and action pairs. Select Star's page lists seven tools, all search, get or traverse operations, and the directory records its access as read-only. ## Some servers ship writes in the default tool list On other servers the write tools are present from the first connection and the control is a role, a header or a tier. Per Airbyte's docs, read-only mode is enabled with the X-MCP-Readonly-Mode header or ?read-only=true, which shows only read-only tools and hides run and cancel sync tools. Per Monte Carlo's docs, an x-mcp-readonly header hides write tools such as update_alert and the create_or_update monitor tools, and blocks calls to them. SYNQ's docs describe four permission tiers (read-only, write, governance, deploy), and write tools do not appear in the client if the user granted only read-only access. Atlan's security page says an optional read-only mode removes every write, admin and lifecycle tool while keeping search and lineage, and that it is requested from Atlan support. Looker's tools in MCP Toolbox include creating or deleting LookML project files and switching Looker into dev mode, so per the Toolbox index access is not read-only by default. Omni lists putApp, patchApp and publishDraft; per Omni's docs, edits land on a shared draft until publishDraft runs, and Organization Admins can enable or disable tool capabilities individually. ## Confirmation is documented in five different forms The Redshift README says each write in read-write mode prompts for confirmation through MCP elicitation unless UNSAFE_SKIP_WRITE_CONFIRMATION is set. Collibra's README describes a confirm checkpoint for destructive data quality tools, where confirm=false is the read-only default. Per Euno's docs, deletes need confirm=true and write tools such as catalog creates ask for client confirmation. Per Atlan's tools page, tools that change anything propose first and wait for approval before writing. Per Tableau's docs, the mutating flow tools (run-flow, run-flow-task, cancel-flow-run) are preview-and-confirm tools, and delete-content is admin-only and two-phase. Microsoft's README describes --readonly as a safe mode that prevents any write to the semantic model, with --readwrite as the default and a confirmation prompt. Fivetran's README is explicit that write and delete confirmation is advisory only and that the server does not enforce it. The Sifflet server has two state-changing tools, open_incident_by_id and close_incident_by_id, and the README and docs page describe no per-call permission prompt (none found on 10 Nov 2026). Secoda's docs list run_sql and do not describe a read-only mode (none found on 10 Nov 2026). ## Tool annotations are published by some servers and treated as hints by others MCP tool annotations such as readOnlyHint and destructiveHint tell a client how a tool behaves. Per the docs for [the Cube MCP server](/servers/cube-mcp/), nine tools are annotated destructive and that clients honoring annotations always ask before running them. The same page says chat is annotated read-only but runs the Analytics Chat agent, which can create reports or delete folders without confirmation, so the annotation understates what it can do. In Tableau's source, query-datasource is annotated readOnlyHint true and destructiveHint false. In ThoughtSpot's tool definitions, search_objects and get_data carry the same pair of values. In the MCP Toolbox BigQuery docs, the source options readOnly and writeMode set the behavior: with readOnly true and writeMode blocked, only SELECT is allowed and the tool is annotated readOnlyHint true. Snowflake's docs state that tool annotations are client hints and not security controls, and that RBAC is the control. The Toolbox docs list a Read-Only Tools page covering engine-level protocol locks, tool suppression and MCP annotations. Whether a client acts on an annotation is a property of the client, so the annotation tells an engineer what the publisher intended and not what the client will enforce. ## Write scope can be narrowed below the server level Several servers limit writes by place instead of by tool. Per Cube's docs, writeDataModelFile and deleteDataModelFile reject any branch that is not a dev branch, so file writes never touch production, and a merge to the default branch goes through a tool that prompts for confirmation. Per Microsoft's Learn page, with only Build permission the Power BI server can only run DAX queries, and Write permission is needed to change model objects. Per Airbyte's tools page, permanent deletes only work when the resource name contains delete-me or deleteme. Per Hex's docs, editing a cell needs Can Edit or Full Access on that project, even for Admins. Per Euno's docs, context benchmark tools are accepted only for an account admin persona. BigQuery's allowedDatasets option restricts the Toolbox tools to listed datasets, and BIGQUERY_MAXIMUM_BYTES_BILLED caps bytes scanned per query. ## The control that holds sits where the credential is checked A header or flag in a client config stops the tool list from showing a write tool. A role or privilege stops the write at the system that owns the data. Airbyte's docs say headers can only hide tools and cannot grant permissions the user's Airbyte role lacks. Snowflake's docs list distinct privileges: USAGE on the MCP server plus SELECT on the semantic view, or USAGE on the Cortex Search service, agent or UDF. The table puts several servers' documented controls side by side. It records what each publisher wrote and does not grade the controls. | Server | Write tools shipped | Documented control | | --- | --- | --- | | Airbyte MCP (hosted) | deploy, update, delete, run sync | Read-only header or query parameter; role limits | | Amazon Redshift MCP Server (AWS Labs) | execute_query in read-write mode | ACCESS_MODE; elicitation prompt per write | | Collibra MCP Server | edit_asset, create_asset, 18 data quality tools | Data quality tools off by default; confirm flag | | Cube MCP server | report, workbook, data model file tools | Nine destructive annotations; dev branch only | | DataHub MCP Server | tags, terms, owners, domains | Off unless TOOLS_IS_MUTATION_ENABLED=true | | Fivetran MCP Server | connections_write, connections_delete | FIVETRAN_SCOPE; denylist; confirmation advisory | | Select Star MCP Server | none found on 10 Nov 2026 | Seven search, get and traverse tools | | SYNQ MCP | create, delete and save tools | Four permission tiers chosen at consent | ## Sources - AWS Labs Redshift MCP README: https://raw.githubusercontent.com/awslabs/mcp/main/src/redshift-mcp-server/README.md (read 10 Nov 2026) - DataHub MCP README: https://raw.githubusercontent.com/acryldata/mcp-server-datahub/main/README.md (read 10 Nov 2026) - Community mcp-snowflake-server README: https://raw.githubusercontent.com/isaacwasserman/mcp-snowflake-server/main/README.md (read 10 Nov 2026) - Collibra MCP README: https://github.com/collibra/chip/blob/main/README.md (read 10 Nov 2026) - Fivetran MCP README: https://raw.githubusercontent.com/fivetran/fivetran-mcp/main/README.md (read 10 Nov 2026) - Select Star MCP docs: https://docs.selectstar.com/features/mcp-server (read 10 Nov 2026) - Airbyte MCP advanced configuration: https://docs.airbyte.com/platform/airbyte-mcp/advanced-configuration.md (read 10 Nov 2026) - Airbyte MCP tools: https://docs.airbyte.com/platform/airbyte-mcp/tools.md (read 10 Nov 2026) - Monte Carlo MCP reference: https://docs.getmontecarlo.com/docs/mcp-server-reference-and-operations (read 10 Nov 2026) - SYNQ MCP docs: https://docs.synq.io/scout/mcp.md (read 10 Nov 2026) - Atlan MCP security page: https://docs.atlan.com/product/capabilities/atlan-ai/references/mcp-security (read 10 Nov 2026) - Atlan MCP tools page: https://docs.atlan.com/product/capabilities/atlan-ai/references/mcp-tools (read 10 Nov 2026) - MCP Toolbox llms.txt index: https://mcp-toolbox.dev/llms.txt (read 10 Nov 2026) - Omni MCP tools: https://docs.omni.co/ai/mcp/tools.md (read 10 Nov 2026) - Euno MCP setup docs: https://docs.euno.ai/ai/mcp-setup.md (read 10 Nov 2026) - Tableau MCP docs intro: https://tableau.github.io/tableau-mcp/docs/intro (read 10 Nov 2026) - Power BI Modeling MCP README: https://raw.githubusercontent.com/microsoft/powerbi-modeling-mcp/main/README.md (read 10 Nov 2026) - Microsoft Learn Power BI authoring MCP: https://learn.microsoft.com/en-us/power-bi/developer/mcp/power-bi-authoring-mcp (read 10 Nov 2026) - Sifflet MCP README: https://github.com/siffletdata/sifflet-mcp/blob/main/README.md (read 10 Nov 2026) - Secoda MCP docs: https://docs.secoda.co/features/ai-assistant/secoda-mcp-server.md (read 10 Nov 2026) - Cube MCP server docs: https://docs.cube.dev/docs/integrations/mcp-server.md (read 10 Nov 2026) - Tableau queryDatasource source: https://raw.githubusercontent.com/tableau/tableau-mcp/main/src/tools/web/queryDatasource/queryDatasource.ts (read 10 Nov 2026) - ThoughtSpot tool definitions: https://raw.githubusercontent.com/thoughtspot/mcp-server/main/src/servers/tool-definitions.ts (read 10 Nov 2026) - MCP Toolbox BigQuery source: https://raw.githubusercontent.com/googleapis/mcp-toolbox/main/docs/en/integrations/bigquery/source.md (read 10 Nov 2026) - MCP Toolbox BigQuery prebuilt config: https://raw.githubusercontent.com/googleapis/mcp-toolbox/main/docs/en/integrations/bigquery/prebuilt-configs/bigquery.md (read 10 Nov 2026) - Snowflake-managed MCP docs: https://docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-mcp (read 10 Nov 2026) - Hex MCP server docs: https://learn.hex.tech/docs/api-integrations/mcp-server (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/guides/read-versus-write/ --- # Whether an agent sees what the asking person may see depends on whether the server runs calls as that person or as one shared credential GUIDE, read 10 Nov 2026 **The short answer.** Atlan, Cube, Hex, Omni and Tableau document calls that run as the signed-in user, while Fivetran uses one API key and secret, the Secoda wrapper one API_TOKEN, and Select Star one API token. Cube applies row-level security, Euno and OpenMetadata use personas, and Looker passes the caller's OAuth token only when use_client_oauth is switched on. ## OAuth on behalf of the user ties each call to a person When a server completes OAuth for the person at the keyboard, the vendor can apply that person's permissions to every tool call. Per the docs for [the Atlan MCP server](/servers/atlan-mcp/), OAuth makes each tool call run as the signed-in user, so tools run with their own permissions and actions are attributed to them. Per the docs for [the Cube MCP server](/servers/cube-mcp/), every tool runs as the authenticated user, and queries respect Cube permissions including row-level security. Per Omni's docs, MCP tools run with the permissions of a specific Omni user, so results include only what that user can access. Per Sigma's help page, the server inherits the user's Sigma account permissions, and access needs the 'Use Sigma MCP with OAuth' account-type permission plus at least Can view on documents or Can use on connections. Per Tableau's README, the hosted server uses OAuth 2.1 so each user signs in as themselves and per-user Tableau permissions are enforced. Google documents OAuth 2.0 with IAM for the BigQuery remote server, so access follows the authenticated Google Cloud identity. Per Databricks, the dbsql MCP runs SQL using the caller's Unity Catalog and warehouse permissions, so results are scoped to the asking user. ## Service accounts and shared keys scope the agent, not the person A shared credential gives every asker the same view. Per the README of [the Fivetran MCP Server](/servers/fivetran-mcp/), the server uses one Fivetran API key and secret, so agent actions run with that credential's scope instead of the asking user's identity, and OAuth resource-server support in streamable-http mode is not yet functional. Secoda's docs say the open-source wrapper authenticates with a single API_TOKEN, one shared token per server process, while the Secoda-hosted endpoint verifies the user's workspace and AI permissions. Select Star's page describes an API token sent as a Bearer header. No per-user OAuth is described, and the page does not say results are limited to the asking user's permissions (none found on 10 Nov 2026). Per Collibra's README, stdio mode uses one server-wide credential, while HTTP mode advises that MCP clients supply their own Basic Auth headers per request. Alation's guide describes the service-account route: the agent gets the data and features allowed by the role assigned to the service account. Per OpenMetadata's docs, unattended agents should use a Bot service account with a role-based access policy, kept separate from personal logins and auditable. Per Airbyte's tools page, direct reads use the credentials stored in the connector rather than the asking user, so source-side access is not per-user even though the tool calls are. ## Some servers offer both and the default decides the outcome Looker's source page says the Toolbox can pass the caller's OAuth token instead of one service credential via use_client_oauth, which defaults to false. With the default, identity is the API user configured with client_id and client_secret. The same Toolbox documents useClientOAuth for BigQuery, which forwards the client's OAuth token so calls run as the asking user, and impersonateServiceAccount for service account impersonation. Omni lets an Organization API key run tools as another user through the X-MCP-User-ID header. Cube's API-key tokens can act as the key's owner, as another user, or be limited to specific tools, and last up to one hour with no refresh token. Tableau's self-hosted server offers PAT (the default), direct-trust, OAuth and passthrough authentication, per its configuration docs. DataHub splits by transport. Per its README, the Docker HTTP deployment has each MCP client supply its own DataHub token, so permissions and audit identity are preserved per user, and the HTTP entry point refuses to start if a shared server-wide token is configured. Over stdio the README uses DATAHUB_GMS_URL and DATAHUB_GMS_TOKEN. ## Personas narrow what an agent is shown A persona is a named profile of what a role should see. Per Euno's docs, the server uses the active persona, and if none can be identified the user is asked to choose; the docs say the active persona scopes data and actions, and list_personas, current_persona and switch_persona are among the published tools. Per Atlan's security page, guests and members are scoped to their persona, and MCP narrows but never widens what a user's login permits. OpenMetadata's reference lists get_persona_context and find_context, which serve shared AI context documents curated per persona, and get_user_context returns the caller's identity, roles, teams, domains and active persona, with no parameter to look up another user. These tools shape context. The access boundary in each case is the platform's own policy, as the vendors describe it. ## Row-level security and role gates appear in the semantic and BI servers Cube's docs name row-level security for MCP queries. Microsoft's overview says Fabric IQ, its consumption server, answers questions over semantic models and respects permissions, row-level security and object-level security. Microsoft's authoring server goes the other way: its tools manage security roles with row-level security inside the model, per the README. Role gates are more common. Per Monte Carlo's docs, an Editor role or above is required to use the server. Hex requires the Explorer role for knowledge tools and Editor or above for project editing tools. Cube says a Viewer role or higher is needed and that data model editing tools show only for roles allowed to edit the model, Admin and Developer by default. Alation's remote server lists tools dynamically and filters out tools the current user cannot use. Snowflake's docs say access to the MCP server does not give access to tools, since privileges are granted per tool. ## The documented identity of each server, side by side The table records the identity each publisher says a call runs as. The sources for each row are listed at the end of the guide. Where a server offers more than one route, both are shown. A row that reads none found means the page read on 10 Nov 2026 did not state how results are scoped. Per-user scoping also depends on the client and on how the customer configures roles, so the vendor's statement is the starting point for a test and not the end of it. | Server | Identity the call runs as | Per-user scoping documented | | --- | --- | --- | | Atlan MCP (hosted) | Signed-in user via OAuth, or API key as a shared identity | Yes, with OAuth | | BigQuery remote MCP server | Authenticated Google Cloud identity via OAuth 2.0 and IAM | Yes | | Collibra MCP Server | Server-wide credential over stdio; client Basic Auth headers over HTTP | Per request over HTTP | | Cube MCP server | Authenticated user, or API-key owner or chosen user | Yes, with row-level security | | Fivetran MCP Server | One API key and secret | Not documented | | Looker tools in MCP Toolbox | API user, or caller's OAuth token if use_client_oauth is true | Opt-in | | Select Star MCP Server | API token as Bearer header | none found on 10 Nov 2026 | | Tableau MCP | Hosted: each user's Tableau Cloud identity; self-hosted: PAT by default | Hosted only | ## Sources - Atlan remote MCP overview: https://docs.atlan.com/product/capabilities/atlan-ai/how-tos/remote-mcp-overview (read 10 Nov 2026) - Atlan MCP security page: https://docs.atlan.com/product/capabilities/atlan-ai/references/mcp-security (read 10 Nov 2026) - Cube MCP server docs: https://docs.cube.dev/docs/integrations/mcp-server.md (read 10 Nov 2026) - Omni MCP tools: https://docs.omni.co/ai/mcp/tools.md (read 10 Nov 2026) - Sigma MCP server help page: https://help.sigmacomputing.com/docs/use-sigma-mcp-server (read 10 Nov 2026) - Tableau MCP repository: https://github.com/tableau/tableau-mcp (read 10 Nov 2026) - Tableau MCP environment variables: https://tableau.github.io/tableau-mcp/docs/configuration/mcp-config/env-vars (read 10 Nov 2026) - BigQuery remote MCP guide: https://docs.cloud.google.com/bigquery/docs/use-bigquery-mcp (read 10 Nov 2026) - Databricks managed MCP page: https://docs.databricks.com/aws/en/generative-ai/mcp/managed-mcp (read 10 Nov 2026) - Fivetran MCP README: https://raw.githubusercontent.com/fivetran/fivetran-mcp/main/README.md (read 10 Nov 2026) - Secoda MCP docs: https://docs.secoda.co/features/ai-assistant/secoda-mcp-server.md (read 10 Nov 2026) - Select Star MCP docs: https://docs.selectstar.com/features/mcp-server (read 10 Nov 2026) - Collibra MCP README: https://github.com/collibra/chip/blob/main/README.md (read 10 Nov 2026) - Alation authentication guide: https://github.com/Alation/alation-ai-agent-sdk/blob/main/guides/authentication.md (read 10 Nov 2026) - Alation agent SDK README: https://github.com/Alation/alation-ai-agent-sdk/blob/main/README.md (read 10 Nov 2026) - OpenMetadata MCP guide: https://docs.open-metadata.org/v2.0.x/how-to-guides/mcp.md (read 10 Nov 2026) - OpenMetadata MCP reference: https://docs.open-metadata.org/v2.0.x/how-to-guides/mcp/reference.md (read 10 Nov 2026) - Airbyte MCP tools: https://docs.airbyte.com/platform/airbyte-mcp/tools.md (read 10 Nov 2026) - MCP Toolbox Looker source: https://mcp-toolbox.dev/integrations/looker/source/ (read 10 Nov 2026) - MCP Toolbox BigQuery source: https://raw.githubusercontent.com/googleapis/mcp-toolbox/main/docs/en/integrations/bigquery/source.md (read 10 Nov 2026) - DataHub MCP README: https://raw.githubusercontent.com/acryldata/mcp-server-datahub/main/README.md (read 10 Nov 2026) - Euno MCP setup docs: https://docs.euno.ai/ai/mcp-setup.md (read 10 Nov 2026) - Microsoft Learn MCP servers overview: https://learn.microsoft.com/en-us/power-bi/developer/mcp/mcp-servers-overview (read 10 Nov 2026) - Power BI Modeling MCP README: https://raw.githubusercontent.com/microsoft/powerbi-modeling-mcp/main/README.md (read 10 Nov 2026) - Monte Carlo MCP server docs: https://docs.getmontecarlo.com/docs/mcp-server (read 10 Nov 2026) - Hex MCP server docs: https://learn.hex.tech/docs/api-integrations/mcp-server (read 10 Nov 2026) - Snowflake-managed MCP docs: https://docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-mcp (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/guides/scoping-context-to-the-person-asking/ --- # Every tool a server lists is described to the model on connection, so the number of tools a server publishes is the first size signal to read GUIDE, read 10 Nov 2026 **The short answer.** A client that connects to an MCP server loads each tool's name, description and input schema into the model's context. SYNQ's docs list 134 tools, Euno's table 56, Looker's Toolbox index 55, dbt 52, Airbyte 50 and Collibra 49, while Select Star lists 7 and Databricks 4. Counts differ in what they count. ## Tool definitions occupy the context window before the first question An MCP client asks the server for its tool list and gives each tool's name, description and input schema to the model, so the model can choose among them. Those definitions sit in the context window next to the system prompt, the conversation and any query results. A server with many tools, or tools with long descriptions and nested schemas, uses more of that window before the person has typed anything. The directory's [Context Weight Index](/context-weight/) measures that size in tokens, and its [methodology](/context-weight/methodology/) page says how. This guide quotes no token figures. It covers the other signal, which any engineer can read from a vendor page in a minute: how many tools the server lists, and what the vendor says about narrowing the list. Counts also move. Euno's October 2026 release notes add pivot and multi-count aggregation tools through MCP, so a count read on one date describes that date. The directory keeps the read date with each count for that reason. ## Tool counts range from a handful to over a hundred The counts below come from the tool lists recorded in the directory on 10 Nov 2026. SYNQ's docs list 134 tool names. Euno's published table names 56 tools, with names and descriptions only and no parameter schemas. The Looker tools in MCP Toolbox number 55 looker-* entries in the Toolbox llms.txt index. [The dbt MCP Server](/servers/dbt-mcp/) README lists 52, Airbyte's tools page lists 50, and Collibra's README lists 49. Atlan's tools page states 41 tools across nine categories. Tableau's list holds 40 names, with some gated by SKU and, per Tableau's docs, the knowledge context tools on Tableau+ only. Cube's tables list 30, Omni's page 26, the Power BI authoring server 21, ThoughtSpot's definitions 20, Hex 18, Monte Carlo 15 and OpenMetadata 15. Alation lists 12, Sifflet 10, BigQuery's remote server and the Toolbox BigQuery config 9 each, the Redshift server and the community mcp-snowflake-server 8 each, Select Star 7, Snowflake's managed server 6 tool types and Databricks 4 system.ai MCPs. ## A count does not measure size by itself Counts are comparable only if they count the same thing. Fivetran's README describes two discovery tools plus one execution tool per allowed resource and action pair, generated from the Fivetran OpenAPI spec, so the count depends on the scope the key allows. Snowflake's managed server is a Snowflake object created with CREATE MCP SERVER, so the tool list belongs to the account's specification and not to a vendor page. The Databricks Labs Unity Catalog server infers its tools at startup from the functions and vector search indexes in a given schema, per its README. Sigma's help page describes capabilities (search, describe, query, build workbooks, screenshots, call agents) and publishes no tool names, so no count exists for it (none found on 10 Nov 2026). The Power BI authoring server lists 21 names that each act as an operation group; Microsoft's README describes, for example, database_operations covering TMDL import and export and deployment to Fabric, so one name can carry several actions. Descriptions and schemas matter as much as count. A tool list of 8 with long parameter schemas can use more window than 20 short ones. That is why the Context Weight Index reads the definition text and counts tokens, and why this guide does not order servers by count. The same capability can appear under two names. The dbt list carries get_lineage and get_lineage_dev, and get_node_details and get_node_details_dev, so its count includes pairs for platform and local use. Tableau gates query-knowledge-context, manage-knowledge-context and inspect-knowledge-context to Tableau+, so the list a user sees depends on the SKU. Collibra's data quality group of 18 tools is off by default behind a --data-quality flag. ## Several servers let the client or the admin shrink the list The tool list an agent receives is not always the full list. Tableau's docs describe INCLUDE_TOOLS and EXCLUDE_TOOLS environment variables, taking comma-separated tool or group names. The dbt Labs repository's .env.example shows tool groups enabled or disabled by environment variables. Collibra's README says specific tools can be enabled or disabled through flags, environment variables or mcp.yaml, and by default all are enabled. Alation's README describes an explicit enabled-tools list, set by CLI argument or ALATION_ENABLED_TOOLS. Per Monte Carlo's docs, per-request x-mcp-tools allow-lists narrow tools and are enforced on call, not only on listing. Airbyte's read-only mode shows only read-only tools. Per Omni's docs, Organization Admins can enable or disable tool capabilities on the MCP settings page, and a user-scoped key lists a tool only if the member holds the permission it needs. Role-based filtering does the same through permissions. SYNQ's write tools do not appear if the user granted only read-only access at consent. Cube's data model editing tools show only for roles allowed to edit the model. Alation's remote server filters out tools the current user cannot use. ## Some servers add tools for the agent to read its own context A few lists carry tools that exist to orient the agent. Euno's list includes load_account_settings, list_personas, current_persona, switch_persona and load_skill. SYNQ's iam tool reports the caller's workspace, granted scopes and user. The OpenMetadata reference lists get_user_context. DataHub's optional get_me tool, enabled with TOOLS_IS_USER_ENABLED=true, returns the authenticated user's profile and group memberships. These are small in count and useful for scoping, and they are also part of the definition payload. A server that lists orientation tools and 50 domain tools carries both. ## Counts side by side for a sample of servers The table lists a sample in alphabetical order with what the count refers to. It states numbers and does not call any of them high or low. A count of tool names is only the first signal. It says nothing about description length or schema depth, which is what the Context Weight Index measures. | Server | Tool names listed | What is counted | | --- | --- | --- | | Airbyte MCP (hosted) | 50 | Names on the Airbyte MCP tools page | | Atlan MCP (hosted) | 41 | Tools page states 41 across nine categories | | Collibra MCP Server | 49 | README tool list, including 18 data quality tools off by default | | dbt MCP Server | 52 | README tool list across tool groups | | Euno MCP | 56 | Published table of names and descriptions | | Fivetran MCP Server | 2 plus generated | Discovery tools plus one execution tool per allowed pair | | Select Star MCP Server | 7 | Tools listed on the docs page | | Sigma MCP server | none found on 10 Nov 2026 | Docs publish capabilities, no tool names | | SYNQ MCP | 134 | Names on the SYNQ MCP docs page | ## Sources - SYNQ MCP docs: https://docs.synq.io/scout/mcp.md (read 10 Nov 2026) - Euno MCP setup docs: https://docs.euno.ai/ai/mcp-setup.md (read 10 Nov 2026) - MCP Toolbox llms.txt index: https://mcp-toolbox.dev/llms.txt (read 10 Nov 2026) - dbt MCP README: https://raw.githubusercontent.com/dbt-labs/dbt-mcp/main/README.md (read 10 Nov 2026) - dbt MCP .env.example: https://raw.githubusercontent.com/dbt-labs/dbt-mcp/main/.env.example (read 10 Nov 2026) - Airbyte MCP tools: https://docs.airbyte.com/platform/airbyte-mcp/tools.md (read 10 Nov 2026) - Collibra MCP README: https://github.com/collibra/chip/blob/main/README.md (read 10 Nov 2026) - Atlan MCP tools page: https://docs.atlan.com/product/capabilities/atlan-ai/references/mcp-tools (read 10 Nov 2026) - Tableau MCP docs intro: https://tableau.github.io/tableau-mcp/docs/intro (read 10 Nov 2026) - Tableau MCP environment variables: https://tableau.github.io/tableau-mcp/docs/configuration/mcp-config/env-vars (read 10 Nov 2026) - Cube MCP server docs: https://docs.cube.dev/docs/integrations/mcp-server.md (read 10 Nov 2026) - Omni MCP tools: https://docs.omni.co/ai/mcp/tools.md (read 10 Nov 2026) - Power BI Modeling MCP README: https://raw.githubusercontent.com/microsoft/powerbi-modeling-mcp/main/README.md (read 10 Nov 2026) - ThoughtSpot tool definitions: https://raw.githubusercontent.com/thoughtspot/mcp-server/main/src/servers/tool-definitions.ts (read 10 Nov 2026) - Hex MCP server docs: https://learn.hex.tech/docs/api-integrations/mcp-server (read 10 Nov 2026) - Monte Carlo MCP reference: https://docs.getmontecarlo.com/docs/mcp-server-reference-and-operations (read 10 Nov 2026) - OpenMetadata MCP reference: https://docs.open-metadata.org/v2.0.x/how-to-guides/mcp/reference.md (read 10 Nov 2026) - Alation agent SDK README: https://github.com/Alation/alation-ai-agent-sdk/blob/main/README.md (read 10 Nov 2026) - Sifflet MCP README: https://github.com/siffletdata/sifflet-mcp/blob/main/README.md (read 10 Nov 2026) - BigQuery MCP tools reference: https://docs.cloud.google.com/bigquery/docs/reference/mcp (read 10 Nov 2026) - MCP Toolbox BigQuery prebuilt config: https://raw.githubusercontent.com/googleapis/mcp-toolbox/main/docs/en/integrations/bigquery/prebuilt-configs/bigquery.md (read 10 Nov 2026) - AWS Labs Redshift MCP README: https://raw.githubusercontent.com/awslabs/mcp/main/src/redshift-mcp-server/README.md (read 10 Nov 2026) - Community mcp-snowflake-server README: https://raw.githubusercontent.com/isaacwasserman/mcp-snowflake-server/main/README.md (read 10 Nov 2026) - Select Star MCP docs: https://docs.selectstar.com/features/mcp-server (read 10 Nov 2026) - Snowflake-managed MCP docs: https://docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-mcp (read 10 Nov 2026) - Databricks managed MCP page: https://docs.databricks.com/aws/en/generative-ai/mcp/managed-mcp (read 10 Nov 2026) - Fivetran MCP README: https://raw.githubusercontent.com/fivetran/fivetran-mcp/main/README.md (read 10 Nov 2026) - Databricks Labs MCP README: https://raw.githubusercontent.com/databrickslabs/mcp/master/README.md (read 10 Nov 2026) - Sigma MCP server help page: https://help.sigmacomputing.com/docs/use-sigma-mcp-server (read 10 Nov 2026) - DataHub MCP README: https://raw.githubusercontent.com/acryldata/mcp-server-datahub/main/README.md (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/guides/why-tool-definition-size-matters/ --- # The auth route a server documents decides who is on the audit trail, and few servers document an audit log for MCP calls at all GUIDE, read 10 Nov 2026 **The short answer.** Auth across the directory spans OAuth, personal access tokens, API keys, Basic Auth and cloud credentials. Collibra uses a username and password, Redshift uses AWS credentials, and Monte Carlo accepts only MCP-specific API keys. Audit evidence is thinner: Atlan logs each call with request ID, BigQuery cites centralized audit logging, and Sifflet shows none found. ## OAuth is the documented route on most hosted servers Hosted servers lean on OAuth because the vendor already holds the identity. Airbyte documents OAuth over Streamable HTTP as the signed-in user, or an Airbyte application for non-browser clients. Cube documents OAuth Authorization Code with PKCE under the scope mcp-agent-access. Omni documents OAuth 2.1 with PKCE, which creates an MCP OAuth personal access token for the user. Monte Carlo documents OAuth 2.1 with Dynamic Client Registration. Registration is a practical difference. Per Alation's README, the remote MCP server does not support dynamic client registration, so clients must supply a client_id and client_secret. Per ThoughtSpot's README, OAuth works through dynamic client registration or a manually registered client ID and secret. Atlan documents OAuth per user, and Hex documents OAuth authorization to the user's workspace. Consent screens and settings differ. SYNQ's docs say the consent screen offers only what the user's role allows, and admins can switch off OAuth access to the API for third-party clients, including MCP, with the Disable user API access setting. Euno documents OAuth sign-in by default and an API-key setup for clients that cannot do interactive sign-in. Select Star's troubleshooting note says the token must carry a Bearer prefix. ## Personal access tokens, API keys and Basic Auth cover the rest Where OAuth is absent or optional, the server takes a token. Select Star documents an API token sent as an Authorization Bearer header. [Sifflet](/servers/sifflet-mcp/) documents an API token and backend URL as environment variables, with the Viewer role enough for most tools and the Editor role needed for get_monitor_code_by_description. Fivetran documents an API key and secret generated in its dashboard. DataHub documents a personal access token as an Authorization Bearer header over HTTP, and rejects tokens in the query string. Per the docs for [the Monte Carlo MCP Server](/servers/monte-carlo-mcp/), API-key access uses MCP-server-specific scoped keys and that standard API keys do not work; the key is passed as id:secret in a Bearer header through mcp-remote. Atlan accepts an API key as a Bearer token for a shared service identity. Sigma's API-key access token, scope mcp:access, is labeled Beta. [Collibra](/servers/collibra-mcp/) documents a username and password, either server-wide over stdio or as client-supplied Basic Auth headers over HTTP. Cube exchanges an API key for a token, with client_id cube-api-key, and the token lasts up to one hour with no refresh token. OpenMetadata accepts a personal access token or a Bot JWT as an Authorization Bearer header, alongside OAuth 2.0 through existing SSO. Secoda's local wrapper reads an API_TOKEN environment variable, one shared token per server process, and its docs describe the hosted endpoint as authenticating against workspace and AI permissions. Per Sifflet's README, a Viewer-role token is enough for most tools, which keeps the credential's reach small by role. ## Cloud-native servers use the cloud's own credentials The warehouse and cloud servers inherit credential systems that already exist. The Redshift server uses AWS credentials from the CLI, a profile or environment variables, with IAM permissions for the Redshift Data API. BigQuery's remote server uses OAuth 2.0 with IAM, and the Toolbox BigQuery tools take Application Default Credentials, client OAuth token forwarding or service account impersonation. Microsoft's Power BI authoring server uses Microsoft Entra interactive sign-in by default locally, or a service principal with a client secret or certificate; the hosted preview endpoint uses Entra ID as the signed-in user. Snowflake's managed server uses Snowflake OAuth by default with optional External OAuth, and supports a PAT. The Snowflake Labs server, per its README, supports password or PAT, key pair, OAuth, SSO and MFA through the Snowflake connector. Databricks documents workspace sign-in with Unity Catalog permissions. ## Audit evidence is documented for only some servers Per Atlan's security page, every tool call is logged as a structured event with tool name, client, tenant, request ID and status. Google's BigQuery guide says its remote MCP servers offer centralized audit logging and optional Model Armor protection. Per Sigma's help page, MCP queries can be monitored in the data platform using the query tag "kind":"mcp". Identity-based attribution stands in for a log on some servers. SYNQ's docs say OAuth 2 with PKCE means actions are recorded as the real user with a via MCP badge. DataHub's README says per-user tokens preserve DataHub permissions and audit identity. Atlan's docs say OAuth attributes actions to the user. OpenMetadata's docs describe Bot service accounts as auditable. Other mentions are about adjacent controls. The MCP Toolbox README says OpenTelemetry observability is built in, and Tableau's env-vars page says telemetry defaults to the noop provider. Monte Carlo's AI security page says external security audits and penetration tests happen at least annually and cover MCP surfaces, and its reference lets admins restrict the server by IP with Network Access Control Lists. Those are security practices, not logs of calls. ## Where nothing was found, the directory says so Sifflet's README and docs page describe no audit log or per-call permission prompt (none found on 10 Nov 2026). For Collibra, Fivetran and Select Star the recorded facts include no audit-log statement (none found on 10 Nov 2026). Euno's published tool table lists search_resource_audit_log, a tool that searches a resource audit log; the docs read do not say whether it records MCP calls (none found on 10 Nov 2026). An absence in the recorded facts is not a claim that a log does not exist. It means the page or README read on that date did not describe one, and an engineer who needs a log should ask the vendor and write down the answer. ## Auth routes and audit statements side by side The table lists a sample in alphabetical order. The audit column quotes only what the publisher documents. Several audit statements describe the vendor's platform logs and not a log of MCP calls specifically. The wording in the table follows each publisher's own page. | Server | Auth route documented | Audit statement | | --- | --- | --- | | Atlan MCP (hosted) | OAuth per user; API key as Bearer | Structured event per tool call | | BigQuery remote MCP server | OAuth 2.0 with IAM | Centralized audit logging | | Collibra MCP Server | Username and password, Basic Auth | none found on 10 Nov 2026 | | DataHub MCP Server | Personal access token per request over HTTP | Per-user audit identity preserved | | Monte Carlo MCP Server | OAuth 2.1 with DCR; MCP-specific API key | Annual external security audits; no call log found on 10 Nov 2026 | | Redshift MCP Server (AWS Labs) | AWS credentials with IAM | none found on 10 Nov 2026 | | Sifflet MCP Server | API token, environment variables | none found on 10 Nov 2026 | | Sigma MCP server | OAuth; API-key token (Beta) | Query tag kind:mcp in the data platform | | SYNQ MCP | OAuth 2 with PKCE; workspace token | Actions recorded as the user with a via MCP badge | ## Sources - Airbyte MCP overview: https://docs.airbyte.com/platform/airbyte-mcp.md (read 10 Nov 2026) - Cube MCP server docs: https://docs.cube.dev/docs/integrations/mcp-server.md (read 10 Nov 2026) - Omni MCP authentication: https://docs.omni.co/ai/mcp/authentication.md (read 10 Nov 2026) - Monte Carlo MCP server docs: https://docs.getmontecarlo.com/docs/mcp-server (read 10 Nov 2026) - Monte Carlo AI security page: https://docs.getmontecarlo.com/docs/ai-security (read 10 Nov 2026) - Monte Carlo MCP reference: https://docs.getmontecarlo.com/docs/mcp-server-reference-and-operations (read 10 Nov 2026) - Alation agent SDK README: https://github.com/Alation/alation-ai-agent-sdk/blob/main/README.md (read 10 Nov 2026) - ThoughtSpot MCP README: https://raw.githubusercontent.com/thoughtspot/mcp-server/main/README.md (read 10 Nov 2026) - Atlan MCP security page: https://docs.atlan.com/product/capabilities/atlan-ai/references/mcp-security (read 10 Nov 2026) - Atlan remote MCP overview: https://docs.atlan.com/product/capabilities/atlan-ai/how-tos/remote-mcp-overview (read 10 Nov 2026) - Hex MCP server docs: https://learn.hex.tech/docs/api-integrations/mcp-server (read 10 Nov 2026) - Select Star MCP docs: https://docs.selectstar.com/features/mcp-server (read 10 Nov 2026) - Sifflet MCP README: https://github.com/siffletdata/sifflet-mcp/blob/main/README.md (read 10 Nov 2026) - Fivetran MCP README: https://raw.githubusercontent.com/fivetran/fivetran-mcp/main/README.md (read 10 Nov 2026) - DataHub MCP README: https://raw.githubusercontent.com/acryldata/mcp-server-datahub/main/README.md (read 10 Nov 2026) - Sigma MCP server help page: https://help.sigmacomputing.com/docs/use-sigma-mcp-server (read 10 Nov 2026) - Collibra MCP README: https://github.com/collibra/chip/blob/main/README.md (read 10 Nov 2026) - OpenMetadata MCP guide: https://docs.open-metadata.org/v2.0.x/how-to-guides/mcp.md (read 10 Nov 2026) - AWS Labs Redshift MCP README: https://raw.githubusercontent.com/awslabs/mcp/main/src/redshift-mcp-server/README.md (read 10 Nov 2026) - BigQuery remote MCP guide: https://docs.cloud.google.com/bigquery/docs/use-bigquery-mcp (read 10 Nov 2026) - MCP Toolbox BigQuery source: https://raw.githubusercontent.com/googleapis/mcp-toolbox/main/docs/en/integrations/bigquery/source.md (read 10 Nov 2026) - Microsoft Learn Power BI authoring MCP: https://learn.microsoft.com/en-us/power-bi/developer/mcp/power-bi-authoring-mcp (read 10 Nov 2026) - Power BI Modeling MCP README: https://raw.githubusercontent.com/microsoft/powerbi-modeling-mcp/main/README.md (read 10 Nov 2026) - Snowflake-managed MCP docs: https://docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-mcp (read 10 Nov 2026) - Snowflake Labs MCP README: https://raw.githubusercontent.com/Snowflake-Labs/mcp/main/README.md (read 10 Nov 2026) - Databricks managed MCP page: https://docs.databricks.com/aws/en/generative-ai/mcp/managed-mcp (read 10 Nov 2026) - SYNQ MCP docs: https://docs.synq.io/scout/mcp.md (read 10 Nov 2026) - MCP Toolbox README: https://raw.githubusercontent.com/googleapis/genai-toolbox/main/README.md (read 10 Nov 2026) - Tableau MCP environment variables: https://tableau.github.io/tableau-mcp/docs/configuration/mcp-config/env-vars (read 10 Nov 2026) - Euno MCP setup docs: https://docs.euno.ai/ai/mcp-setup.md (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/guides/auth-and-audit/ --- # Where a server runs, as a local stdio process or a hosted Streamable HTTP endpoint, decides who patches it, who holds the credential and which clients can reach it GUIDE, read 10 Nov 2026 **The short answer.** Three servers run over stdio only (AWS Labs Redshift, Sifflet, the community mcp-snowflake-server), 15 are hosted-only in the directory's records, and 12 offer both. Hosted examples include Airbyte, Atlan, BigQuery and Cube, while Euno and Monte Carlo document the mcp-remote bridge for clients that speak only stdio. ## A stdio server is a process on the engineer's machine A local server starts as a child process of the client and talks over standard input and output. The engineer installs it, holds its configuration and updates it. Three servers in the directory are recorded as stdio-only: the [Amazon Redshift MCP Server](/servers/awslabs-redshift-mcp-server/), the Sifflet MCP Server and the community mcp-snowflake-server. The install paths are specific. The README of [the Sifflet MCP Server](/servers/sifflet-mcp/) documents setup for Cursor and Claude Desktop through uvx sifflet-mcp@latest. The community Snowflake server runs via uvx or uv for Claude Desktop, and its README also documents installing through Smithery. The Redshift README shows a stdio client config. Fivetran's README says its server runs over stdio by default and needs Python 3.10 or later and uv. Tableau's docs say the package can run through npx (Node 22.7.5 or later), Docker, Heroku, or as Node single-executable binaries. Collibra's server is a Go binary distributed as prebuilt releases for Linux, macOS and Windows. ## A hosted server is an endpoint the vendor runs A hosted server needs no local process, and the vendor owns its version. Google documents a global BigQuery endpoint that uses Streamable HTTP, enabled when the BigQuery API is enabled. Atlan hosts its server per tenant, and its older local server is deprecated in favor of the hosted endpoint, per the Atlan README. [Airbyte](/servers/airbyte-mcp/), Hex, Cube and Select Star each document one vendor-run endpoint. SYNQ's docs describe regional hosts for the EU, US and AU. Some hosted servers live inside the customer's account. Per Alation's README, the remote MCP server is served from the customer's own Alation instance under /ai/mcp. Snowflake's managed server is an object created with CREATE MCP SERVER and reached at an account-scoped HTTP URL. Databricks documents an endpoint pattern built from the workspace hostname and the MCP name. Hex single-tenant, EU and HIPAA workspaces use their own host instead of the default one. Hosting changes the client requirement. Airbyte's install page says clients must support remote MCP servers with OAuth and the Streamable HTTP transport. Sigma's configuration table sets the transport to HTTP. OpenMetadata documents Streamable HTTP under MCP spec 26 Mar 2025, and Snowflake's docs state support for protocol revision 25 Nov 2025. Monte Carlo's docs say OAuth lets any HTTP-transport MCP client connect with no API keys, and its connector is listed in the Claude Connectors Directory. Monte Carlo also lets admins restrict the server by IP with Network Access Control Lists, and Sigma's page says organizations with an IP allowlist must allowlist the AI assistant's outbound IPs with scope API. ## Several servers ship both modes with different rules Twelve servers are recorded as offering both transports, and the modes can differ in auth and scope. Collibra's HTTP mode binds to localhost only, and its credentials differ by mode. DataHub's HTTP entry point refuses a shared server token. Fivetran runs as a hosted HTTP server with --transport streamable-http, but its README says OAuth resource-server support in that mode is not yet functional. Microsoft's Power BI page says the local server is generally available and the hosted endpoint is in preview. Per the same page, Desktop, PBIP files, transactions and traces are available only on the local server, and Microsoft suggests not registering both local and remote at once. The Snowflake Labs README lists stdio (default), sse (legacy) and streamable-http. The Databricks Labs README says local use is via stdio and Databricks Apps deployment uses Streamable HTTP. The Toolbox runs as a local stdio server through npx @toolbox-sdk/server with the --stdio flag, per the repository README, and its Looker and BigQuery tools use it. ## The mcp-remote bridge connects stdio-only clients to hosted servers Some clients speak stdio and cannot open a remote connection. The open-source mcp-remote proxy sits between them: the client starts mcp-remote as a local process, and mcp-remote talks to the hosted URL. Per the docs for [the Euno MCP server](/servers/euno-mcp/), stdio use goes through the third-party open-source mcp-remote bridge and not a Euno-published package. Monte Carlo documents its MCP-server-type API key id:secret as a Bearer token via mcp-remote. Select Star's page notes that the mcp-remote proxy is used for Claude Desktop. A bridge adds a local process that holds the token or runs the OAuth flow. For a security review it is one more program to read. Euno's docs also say the Euno MCP server keeps no session state, so connections survive restarts and scaling; that is a statement about the server and not about the bridge. ## Client support limits what a hosted server can reach Each vendor lists the clients it has tested. Cube documents Claude, ChatGPT, Microsoft 365 Copilot, Cursor and Codex. Hex documents Claude Desktop, Claude Code, Cursor, ChatGPT, Codex and Glean. Sigma documents Claude, ChatGPT, Codex, Cortex Code and Cursor. Omni has setup pages for ChatGPT, Claude Code, Claude Desktop, Codex, Cursor, Microsoft Copilot and VS Code. Client quirks appear in the docs. Hex says Claude connector widgets created by its connector expire after 24 hours and file uploads to a conversation are not supported. ThoughtSpot's README lists /mcp, /sse, /token/* and /bearer/* endpoints on its Worker, so one deployment exposes more than one path. Secoda's wrapper is local FastMCP with an API_TOKEN, while its docs also describe a Secoda-hosted endpoint. Euno's docs describe a dedicated Gemini Enterprise setup path alongside the direct, bridge and API-key setups. Atlan's docs list supported clients including Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, ChatGPT, Snowflake, Google ADK, n8n, Microsoft 365 Copilot, Glean and Databricks. ## Transport, location and bridge per server The table records the transport and where the server runs for a sample, in alphabetical order. 'Both' means the publisher documents stdio and an HTTP mode. The Snowflake-managed row describes where the server lives. Neither the managed server nor the hosted rows require the engineer to patch anything locally, while the stdio rows do. | Server | Transport recorded | Where it runs | | --- | --- | --- | | Airbyte MCP (hosted) | Remote, Streamable HTTP | Airbyte-hosted endpoint | | Amazon Redshift MCP Server (AWS Labs) | stdio | Local process | | Atlan MCP (hosted) | Remote | Atlan-hosted per tenant | | BigQuery remote MCP server | Remote, Streamable HTTP | Google-managed global endpoint | | Collibra MCP Server | Both | Local Go binary; HTTP binds to localhost | | Euno MCP | Remote; stdio via mcp-remote | Euno-hosted | | Monte Carlo MCP Server | Remote; API key via mcp-remote | Monte Carlo-hosted | | Sifflet MCP Server | stdio | Local process via uvx | | Snowflake-managed MCP server | Remote | Object in the Snowflake account | ## Sources - AWS Labs Redshift MCP README: https://raw.githubusercontent.com/awslabs/mcp/main/src/redshift-mcp-server/README.md (read 10 Nov 2026) - Sifflet MCP README: https://github.com/siffletdata/sifflet-mcp/blob/main/README.md (read 10 Nov 2026) - Community mcp-snowflake-server README: https://raw.githubusercontent.com/isaacwasserman/mcp-snowflake-server/main/README.md (read 10 Nov 2026) - Fivetran MCP README: https://raw.githubusercontent.com/fivetran/fivetran-mcp/main/README.md (read 10 Nov 2026) - Tableau MCP repository: https://github.com/tableau/tableau-mcp (read 10 Nov 2026) - Collibra MCP README: https://github.com/collibra/chip/blob/main/README.md (read 10 Nov 2026) - BigQuery remote MCP guide: https://docs.cloud.google.com/bigquery/docs/use-bigquery-mcp (read 10 Nov 2026) - Atlan agent-toolkit MCP README: https://github.com/atlanhq/agent-toolkit/blob/main/modelcontextprotocol/README.md (read 10 Nov 2026) - Atlan MCP security page: https://docs.atlan.com/product/capabilities/atlan-ai/references/mcp-security (read 10 Nov 2026) - SYNQ MCP docs: https://docs.synq.io/scout/mcp.md (read 10 Nov 2026) - Alation agent SDK README: https://github.com/Alation/alation-ai-agent-sdk/blob/main/README.md (read 10 Nov 2026) - Snowflake-managed MCP docs: https://docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-mcp (read 10 Nov 2026) - Databricks managed MCP page: https://docs.databricks.com/aws/en/generative-ai/mcp/managed-mcp (read 10 Nov 2026) - Hex MCP server docs: https://learn.hex.tech/docs/api-integrations/mcp-server (read 10 Nov 2026) - Airbyte MCP install: https://docs.airbyte.com/platform/airbyte-mcp/install.md (read 10 Nov 2026) - Sigma MCP server help page: https://help.sigmacomputing.com/docs/use-sigma-mcp-server (read 10 Nov 2026) - OpenMetadata MCP connect guide: https://docs.open-metadata.org/v2.0.x/how-to-guides/mcp/connect.md (read 10 Nov 2026) - DataHub MCP README: https://raw.githubusercontent.com/acryldata/mcp-server-datahub/main/README.md (read 10 Nov 2026) - Microsoft Learn Power BI authoring MCP: https://learn.microsoft.com/en-us/power-bi/developer/mcp/power-bi-authoring-mcp (read 10 Nov 2026) - Snowflake Labs MCP README: https://raw.githubusercontent.com/Snowflake-Labs/mcp/main/README.md (read 10 Nov 2026) - Databricks Labs MCP README: https://raw.githubusercontent.com/databrickslabs/mcp/master/README.md (read 10 Nov 2026) - MCP Toolbox repository: https://github.com/googleapis/mcp-toolbox (read 10 Nov 2026) - Euno MCP bridge setup: https://docs.euno.ai/ai/mcp-setup/mcp-bridge-setup.md (read 10 Nov 2026) - Euno MCP setup docs: https://docs.euno.ai/ai/mcp-setup.md (read 10 Nov 2026) - Monte Carlo MCP server docs: https://docs.getmontecarlo.com/docs/mcp-server (read 10 Nov 2026) - Select Star MCP docs: https://docs.selectstar.com/features/mcp-server (read 10 Nov 2026) - Cube MCP server docs: https://docs.cube.dev/docs/integrations/mcp-server.md (read 10 Nov 2026) - Omni docs sitemap: https://docs.omni.co/sitemap.xml (read 10 Nov 2026) - ThoughtSpot MCP README: https://raw.githubusercontent.com/thoughtspot/mcp-server/main/README.md (read 10 Nov 2026) - Secoda MCP docs: https://docs.secoda.co/features/ai-assistant/secoda-mcp-server.md (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/guides/remote-versus-local-servers/ --- # In this directory official means the tool's own vendor publishes the server, and the version, release date and deprecation status are read separately GUIDE, read 10 Nov 2026 **The short answer.** The directory marks 29 of 30 servers official, meaning the data tool's vendor publishes them; one community server is marked unofficial. Maintenance varies: Snowflake Labs' 1.4.2 (15 May 2026) is deprecated, dbt-mcp 2.10.1 shipped 3 Nov 2026, ThoughtSpot's 0.5.0 dates from 6 Jun 2025, and Secoda has no releases. ## Official records who publishes the server, not how it behaves The directory's official flag means the vendor of the data tool publishes the server under its own organization or documentation. Twenty-nine of the 30 servers carry the flag. The one without it is the community mcp-snowflake-server, which its README describes as a community project and not a Snowflake publication, under the GPL-3.0 license per the GitHub API. The flag says nothing about scope, safety or fit. A vendor-published server can be in beta, deprecated or unversioned. The sections below separate the publisher classes the evidence shows: publisher-maintained, Labs and community. Support level is a separate claim when a vendor makes it. The Tableau MCP README carries a badge reading Tableau Supported, and the docs site footer credits Salesforce, Inc. as the copyright holder. No equivalent badge was recorded for most other servers (none found on 10 Nov 2026). Licenses are recorded alongside. Collibra's server is Apache-2.0, Sifflet's is MIT, Secoda's is MIT Python on FastMCP, and DataHub's is Apache-2.0 Python published from the acryldata GitHub organization, which is DataHub's company. Tableau's npm package is Apache-2.0, and ThoughtSpot's npm package is under a ThoughtSpot end user license agreement. Microsoft's README says use is subject to the Power BI Authoring MCP Server EULA and NPX installs require explicit acceptance. ## Publisher-maintained servers differ in how they publish versions Some vendors tag releases. The repository for [the dbt MCP Server](/servers/dbt-mcp/) shows v2.10.1 as latest, with a release timestamp of 3 Nov 2026; PyPI shows the same date. Collibra's repository shows v0.0.67 published 27 Oct 2026, so the server is still at a 0.0.x version. Fivetran's latest release is v0.5.1, timestamped 6 Oct 2026. Microsoft's Power BI authoring server has v1.0.1 on 6 Nov 2026, with v1.0.0 on the npm registry on 25 Sep 2026. Tableau's npm package @tableau/mcp-server has latest 6.0.4, published 9 Nov 2026, under Apache-2.0. Alation's alation-ai-agent-mcp is at 1.0.0, released 26 Oct 2026, which its README calls the first stable release of the SDK. DataHub's mcp-server-datahub is at 0.8.0, uploaded 21 Oct 2026. OpenMetadata's server follows the platform: release 2.0.7 on 27 Oct 2026, with 2.1.0-rc1 also listed. Other vendors publish a service and no version. Airbyte, Cube, Hex, Omni, Sigma, Euno, Monte Carlo, SYNQ and Select Star each document a hosted server with no versioned release (none found on 10 Nov 2026 for each). In those cases the date to record is the docs page's own update stamp where it exists: Monte Carlo's page carries an updatedAt of 2 Nov 2026, Google's BigQuery reference shows Last updated 3 Nov 2026, and Databricks' page shows Nov 9, 2026. ## Labs servers sit between product and experiment Three Labs-branded servers appear. AWS Labs publishes the Redshift server inside its awslabs/mcp repository; PyPI lists awslabs.redshift-mcp-server 0.1.3 uploaded 9 Nov 2026 under the author Amazon Web Services. [Snowflake Labs' repository](/servers/snowflake-labs-mcp/) is marked deprecated and unmaintained in its README, which points users to the Snowflake-managed server; the last release is 1.4.2 on 15 May 2026, titled as the release that starts deprecation. The Databricks Labs repository is the third. Its README calls the Unity Catalog server deprecated in favor of managed MCP servers, no GitHub release entries were read, and the GitHub API shows a last push on 26 Oct 2026 with no archive flag. A developer tools server in the same repository is marked under construction and not yet usable, per the README. Labs here is a publisher label, and the deprecation notices are the vendors' own words. The label alone does not tell an engineer whether a server is maintained, so the README banner and the last release date are the facts to read. ## Deprecation is stated in different places Atlan's local server in atlanhq/agent-toolkit is deprecated in favor of the hosted endpoint per its README; its last GitHub release is v0.3.3 on 17 Feb 2026, and the hosted server has no public version number. Alation removed the metadata-writing tools update_catalog_asset_metadata and check_job_status from its SDK in Jan 2026, per its README, and leaves curation to the REST API. Fivetran's README says the PyPI names fivetran-mcp and mcp-fivetran belong to unrelated projects, so its server is run from the git URL and has no versioned PyPI package. Looker's tools ship in the MCP Toolbox, whose repository moved from genai-toolbox to mcp-toolbox; v1.16.0 was released 5 Nov 2026 after v1.15.0, v1.14.0 and v1.13.1. [Secoda's open-source wrapper](/servers/secoda-mcp/) shows 10 commits, no GitHub releases, and a latest commit on main dated 26 Jan 2026. ThoughtSpot's GitHub releases page shows no releases, and its npm package 0.5.0 was published 6 Jun 2025. Sifflet's PyPI history shows 0.1.5 on 2 Jun 2025, 0.1.6 on 3 Sep 2025 and 0.1.7 on 3 Aug 2026. ## Ownership changes are a separate fact from maintenance Select Star's founder post says Snowflake announced plans to acquire the Select Star platform, dated 24 Nov 2025. The Select Star MCP docs page was read through a web.archive.org snapshot dated 14 Jun 2026 because the docs host did not resolve from the runner on 10 Nov 2026. The directory lists the server and records no release date for it. SYNQ's documentation is titled Coalesce Quality Documentation, and the pages read do not explain the relationship between the names. The directory lists the server as SYNQ MCP and records the title as found. ## Publisher class, version and last release for a sample The table follows alphabetical order and records dates as published. Dates are the release timestamps or upload dates published on the cited pages, and the stamps for hosted servers are page update dates and not releases. Hosted servers with no published version are listed in the text above and not repeated here. | Server | Publisher class | Version | Last release or stamp | | --- | --- | --- | --- | | Alation AI Agent SDK MCP server | Vendor | 1.0.0 | 26 Oct 2026 | | Amazon Redshift MCP Server | AWS Labs | 0.1.3 | 9 Nov 2026 | | Collibra MCP Server | Vendor | v0.0.67 | 27 Oct 2026 | | community mcp-snowflake-server | Community | 0.4.0 | 10 Apr 2025 | | dbt MCP Server | Vendor | 2.10.1 | 3 Nov 2026 | | Secoda MCP | Vendor | no releases | Main branch commit 26 Jan 2026 | | Snowflake Labs MCP Server | Labs, deprecated | 1.4.2 | 15 May 2026 | | Tableau MCP | Vendor | 6.0.4 | 9 Nov 2026 | | ThoughtSpot Spotter MCP Server | Vendor | 0.5.0 | 6 Jun 2025 | ## Sources - Community mcp-snowflake-server README: https://raw.githubusercontent.com/isaacwasserman/mcp-snowflake-server/main/README.md (read 10 Nov 2026) - PyPI mcp-snowflake-server: https://pypi.org/pypi/mcp-snowflake-server/json (read 10 Nov 2026) - Tableau MCP README: https://raw.githubusercontent.com/tableau/tableau-mcp/main/README.md (read 10 Nov 2026) - Tableau MCP docs intro: https://tableau.github.io/tableau-mcp/docs/intro (read 10 Nov 2026) - npm @tableau/mcp-server: https://registry.npmjs.org/@tableau/mcp-server (read 10 Nov 2026) - dbt-mcp GitHub releases: https://github.com/dbt-labs/dbt-mcp/releases (read 10 Nov 2026) - Collibra chip GitHub releases: https://github.com/collibra/chip/releases (read 10 Nov 2026) - Fivetran MCP GitHub releases: https://github.com/fivetran/fivetran-mcp/releases (read 10 Nov 2026) - Power BI Modeling MCP releases: https://github.com/microsoft/powerbi-modeling-mcp/releases (read 10 Nov 2026) - npm @microsoft/powerbi-modeling-mcp: https://registry.npmjs.org/@microsoft/powerbi-modeling-mcp (read 10 Nov 2026) - PyPI alation-ai-agent-mcp: https://pypi.org/pypi/alation-ai-agent-mcp/json (read 10 Nov 2026) - Alation agent SDK README: https://github.com/Alation/alation-ai-agent-sdk/blob/main/README.md (read 10 Nov 2026) - PyPI mcp-server-datahub: https://pypi.org/pypi/mcp-server-datahub/json (read 10 Nov 2026) - OpenMetadata 2.0.7 release: https://github.com/open-metadata/OpenMetadata/releases/tag/2.0.7-release (read 10 Nov 2026) - Monte Carlo MCP server docs: https://docs.getmontecarlo.com/docs/mcp-server (read 10 Nov 2026) - BigQuery MCP tools reference: https://docs.cloud.google.com/bigquery/docs/reference/mcp (read 10 Nov 2026) - Databricks managed MCP page: https://docs.databricks.com/aws/en/generative-ai/mcp/managed-mcp (read 10 Nov 2026) - PyPI awslabs.redshift-mcp-server: https://pypi.org/pypi/awslabs.redshift-mcp-server/json (read 10 Nov 2026) - Snowflake Labs MCP README: https://raw.githubusercontent.com/Snowflake-Labs/mcp/main/README.md (read 10 Nov 2026) - Snowflake Labs MCP releases: https://github.com/Snowflake-Labs/mcp/releases (read 10 Nov 2026) - Databricks Labs MCP README: https://raw.githubusercontent.com/databrickslabs/mcp/master/README.md (read 10 Nov 2026) - Atlan agent-toolkit changelog: https://github.com/atlanhq/agent-toolkit/blob/main/CHANGELOG.md (read 10 Nov 2026) - Atlan agent-toolkit MCP README: https://github.com/atlanhq/agent-toolkit/blob/main/modelcontextprotocol/README.md (read 10 Nov 2026) - Fivetran MCP README: https://raw.githubusercontent.com/fivetran/fivetran-mcp/main/README.md (read 10 Nov 2026) - MCP Toolbox GitHub releases: https://github.com/googleapis/mcp-toolbox/releases (read 10 Nov 2026) - Secoda MCP GitHub releases: https://github.com/secoda/secoda-mcp/releases (read 10 Nov 2026) - Secoda MCP latest commit: https://api.github.com/repos/secoda/secoda-mcp/commits?per_page=1 (read 10 Nov 2026) - ThoughtSpot MCP GitHub releases: https://github.com/thoughtspot/mcp-server/releases (read 10 Nov 2026) - npm @thoughtspot/mcp-server: https://registry.npmjs.org/@thoughtspot/mcp-server (read 10 Nov 2026) - PyPI sifflet-mcp: https://pypi.org/project/sifflet-mcp/ (read 10 Nov 2026) - Select Star post on the Snowflake deal: https://selectstar.com/resources/snowflake-acquire-select-star (read 10 Nov 2026) - Select Star MCP docs (archive snapshot): https://web.archive.org/web/20260614045618/https://docs.selectstar.com/features/mcp-server (read 10 Nov 2026) - SYNQ docs llms.txt: https://docs.synq.io/llms.txt (read 10 Nov 2026) - Power BI Modeling MCP README: https://raw.githubusercontent.com/microsoft/powerbi-modeling-mcp/main/README.md (read 10 Nov 2026) - Collibra MCP README: https://github.com/collibra/chip/blob/main/README.md (read 10 Nov 2026) - DataHub MCP README: https://raw.githubusercontent.com/acryldata/mcp-server-datahub/main/README.md (read 10 Nov 2026) - Sifflet MCP pyproject: https://raw.githubusercontent.com/siffletdata/sifflet-mcp/main/pyproject.toml (read 10 Nov 2026) - Secoda MCP docs: https://docs.secoda.co/features/ai-assistant/secoda-mcp-server.md (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/guides/what-official-means/ --- # Snowflake data can be reached through a Snowflake-managed server, a deprecated Snowflake Labs server or a community server, and their published facts differ on every field GUIDE, read 10 Nov 2026 **The short answer.** Snowflake documents a managed MCP server created with CREATE MCP SERVER; Snowflake Labs' repository, last released as 1.4.2 on 15 May 2026, is marked deprecated; and the community mcp-snowflake-server, 0.4.0 from 10 Apr 2025, runs over stdio. Publisher, transport, auth, tools and write controls are set side by side. ## Three servers, three publishers, three statuses The [Snowflake-managed MCP server](/servers/snowflake-managed-mcp/) is documented by Snowflake as a way for agents to retrieve data from Snowflake accounts without deploying separate infrastructure. The [Snowflake Labs MCP server](/servers/snowflake-labs-mcp/) lives in the Snowflake-Labs/mcp repository, whose README states it is deprecated and no longer maintained and points users to the Snowflake-managed server. The [community mcp-snowflake-server](/servers/isaacwasserman-mcp-snowflake-server/) is a community project, not published by Snowflake, under GPL-3.0 per the GitHub API. On the version fields, the managed server publishes no semantic version or release date; Snowflake's docs state support for MCP protocol revision 25 Nov 2025. Snowflake Labs' latest release, per its releases page and PyPI, is 1.4.2, published 15 May 2026 and titled as the release that starts deprecation. The community server is at 0.4.0 on PyPI, dated 10 Apr 2025, and its GitHub repository was last pushed 7 Oct 2025. ## Where each one runs and how it is installed Snowflake's managed server is a Snowflake object created with CREATE MCP SERVER from a YAML specification and reached at an account-scoped HTTP URL. Nothing is installed on the engineer's machine, and the server is a securable object, so it can be created, replaced and granted on like other Snowflake objects. The docs describe exposing a Cortex Agent as one client-facing tool, in which case the agent chooses among its configured Cortex Analyst, Cortex Search and custom tools for each request. The tools are declared in the specification, with types that Snowflake lists as Cortex Agent, coding agent, Cortex Search, Cortex Analyst, SQL execution and generic UDF or procedure tools. The Snowflake Labs server is installed with uvx or Docker. Its README lists stdio as the default transport, sse as legacy and streamable-http. Tools are configured in a YAML file, and the README lists Cortex Search, Cortex Analyst and Cortex Agent services, an object manager group, a query manager group for SQL execution and a semantic manager group. The community server is run through uvx or uv with stdio for Claude Desktop, and its README documents installing through Smithery. It is recorded as stdio only. The README does not describe a hosted mode, and the directory records no remote endpoint for it (none found on 10 Nov 2026). ## How each one authenticates Per Snowflake's docs, the managed server uses Snowflake OAuth by default with optional External OAuth, supports a PAT, and applies role-based access control per MCP server and per tool. The docs list distinct privileges: USAGE on the MCP server plus SELECT on the semantic view, or USAGE on the Cortex Search service, agent or UDF. Access to the server does not give access to tools. The Snowflake Labs README says authentication uses the Snowflake Python connector and supports password or PAT, key pair, OAuth, SSO and MFA, and that the server honors the RBAC grants of the role used for the connection. The community README documents Snowflake user credentials, a private key path, external browser authentication, or a connections.toml file, and its examples show plain passwords in the config. Scope then follows whatever Snowflake user those credentials belong to; the README describes no per-tool privilege layer of its own. ## What tools each one lists The community server lists eight tools: read_query, write_query, create_table, list_databases, list_schemas, list_tables, describe_table and append_insight. It also exposes a memo://insights resource, and when prefetch is enabled it exposes per-table schema summaries as resources. Per its README, describe_table returns column names, types, nullability, defaults and comments. The managed server's Cortex Analyst tools use semantic views, which hold business definitions, and the docs say the server accepts semantic views only, not semantic models. Snowflake Labs' legacy tooling included semantic view discovery and querying plus Cortex Analyst. None of the three lists a lineage, column-lineage or object-dependency tool (none found on 10 Nov 2026 in each of the three pages read). The three counts are not like for like. The community list is eight named tools, the Snowflake Labs entry is six groups or service types as recorded, and the managed list is six tool types in a specification. They count named functions, configuration groups and specification types, so the table gives each count with what it counts. The community README says append_insight stores insights the agent discovers, which the other two pages do not list as a tool name. ## How each one documents writes and limits The community README says write operations are disabled by default and enabled only with --allow-write, and it supports tool exclusion plus database, schema and table exclusion patterns. The Snowflake Labs README describes a sql_statement_permissions config block that allows or denies SQL statement types one by one. For the managed server, Snowflake's docs warn that tool annotations are client hints and not security controls and that RBAC is the control. The docs also say the managed server is available in government regions except Azure US Gov Virginia and has not completed formal FedRAMP assessment. The read dates are the same for all three, 10 Nov 2026. The Snowflake-managed and Snowflake Labs facts are Snowflake's own statements; the community facts come from the project's README, which is the publisher's own statement as well. Snowflake's docs add that the managed server serves Snowflake data only, with custom tools able to wrap UDFs and stored procedures, and Snowflake Labs' README lists Snowflake as its only coverage. ## The three side by side The table sets the recorded fields next to each other. It states no conclusion, and the order of columns is alphabetical. For the directory's page covering all three, see [MCP servers for Snowflake](/mcp/snowflake/). | Field | Community mcp-snowflake-server | Snowflake Labs MCP Server | Snowflake-managed MCP server | | --- | --- | --- | --- | | Publisher | Community project | Snowflake Labs | Snowflake | | Status | Community; last PyPI release 10 Apr 2025 | Deprecated; 1.4.2 on 15 May 2026 | Documented by Snowflake; MCP revision 25 Nov 2025 | | Transport | stdio | stdio default, sse, streamable-http | Remote, account-scoped HTTP URL | | Auth | User credentials, key path, browser, connections.toml | Connector: password or PAT, key pair, OAuth, SSO, MFA | Snowflake OAuth, External OAuth, PAT | | Tools | 8 named tools | 6 groups or service types, config-driven | 6 tool types from the specification | | Write control | Off unless --allow-write | Per-statement-type permissions | RBAC per MCP server and per tool | | Lineage tool | none found on 10 Nov 2026 | none found on 10 Nov 2026 | none found on 10 Nov 2026 | ## Sources - Snowflake-managed MCP docs: https://docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-mcp (read 10 Nov 2026) - Snowflake Labs MCP README: https://raw.githubusercontent.com/Snowflake-Labs/mcp/main/README.md (read 10 Nov 2026) - Snowflake Labs MCP releases: https://github.com/Snowflake-Labs/mcp/releases (read 10 Nov 2026) - PyPI snowflake-labs-mcp: https://pypi.org/pypi/snowflake-labs-mcp/json (read 10 Nov 2026) - Community mcp-snowflake-server README: https://raw.githubusercontent.com/isaacwasserman/mcp-snowflake-server/main/README.md (read 10 Nov 2026) - PyPI mcp-snowflake-server: https://pypi.org/pypi/mcp-snowflake-server/json (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/guides/three-ways-to-connect-an-agent-to-snowflake/ --- # Catalog, lineage and observability servers describe data, warehouse servers run SQL, and an agent that needs both connects to both GUIDE, read 10 Nov 2026 **The short answer.** Metadata servers such as Atlan, Collibra, DataHub, Monte Carlo and Sifflet describe assets, owners, lineage and incidents. Warehouse servers such as Redshift and BigQuery run SQL. The overlap is small: Atlan's query_assets returns up to 100 rows, Secoda has run_sql, and DataHub's get_dataset_queries fetches SQL without running it. ## Metadata servers answer what a table is, who owns it and what it feeds A metadata server fronts a catalog, a lineage graph or an observability product. Its tools return descriptions, owners, tags, glossary terms, lineage and the state of monitors. Per DataHub's README, agents can discover domains, owners, tags and glossary terms. Per Collibra's README, get_business_term_data traces a business term back to its connected physical data assets. Per Select Star's product page, the server returns descriptions, tags and owners. Observability servers describe pipeline and data health. Monte Carlo's docs say connected agents can investigate alerts, explore assets and lineage, and create and manage monitors. Sifflet's server lists search_incidents, get_incident_scope_by_issue_number and get_downstream_assets_of_asset. Per dbt Labs' README, get_model_health returns run status, test results and upstream source freshness from dbt Platform. Lineage depth is a parameter on some servers. Alation's lineage tool takes max_depth (default 10) and, per Alation's guide, is a BETA feature that Alation support must enable on the instance. OpenMetadata's get_entity_lineage reaches up to 10 hops, with includeColumnLineage defaulting to false so the default answer is table level. Monte Carlo's agent_observability toolset adds get_agent_metadata, get_agent_conversation and get_agent_trace, which describe AI agents and not tables. ## Query servers run statements against an engine A query server's main job is to execute SQL under a database identity. The [Amazon Redshift MCP Server](/servers/awslabs-redshift-mcp-server/) lists execute_query. [BigQuery's remote server](/servers/bigquery-remote-mcp/) lists execute_sql_readonly and execute_sql. The Snowflake-managed server lists a SQL execution tool type, and Databricks documents dbsql. In each, the result depends on the privileges of the calling identity or role, as covered in the guide on [scoping context](/guides/). These servers publish metadata tools too, but their metadata is what the engine itself knows: schemas, tables, columns and comments. BigQuery's get_table_info returns metadata for BigQuery and BigLake objects and lists no glossary or policy tool, per Google's reference. Redshift's README lists no glossary, ownership or classification tool, and retrieve_skill_file returns a Redshift guide covering metadata and SQL. Databricks says its dbsql, sandbox and web_search MCPs require the Unity Gateway beta while Genie One MCP is generally available, and that calls to the Genie Agent MCP bypass Unity Gateway and rely on the permissions of the Genie Agent and its data. Snowflake's managed server requires SELECT on a semantic view for Cortex Analyst tools, a privilege check at the engine and not in a catalog. ## Some metadata servers also execute or retrieve SQL The line between the two kinds is not sharp, so the tools are worth reading by name. Atlan's query_assets runs read-only SQL against a connected warehouse and returns up to 100 rows, and per Atlan's security page tools that execute SQL route it through a read-only validator that permits only SELECT-class queries. Secoda's docs describe run_sql as executing SQL against connected warehouses, and they do not describe a read-only mode (none found on 10 Nov 2026). Alation's guide says its SQL tools need per-data-product data warehouse credentials configured once. DataHub's get_dataset_queries fetches real SQL queries that reference a dataset or column; it returns query history and does not execute a new statement. Select Star's get_asset_queries returns warehouse queries connected to tables or columns. The dbt MCP Server lists execute_sql and text_to_sql alongside its lineage and Semantic Layer tools. ## How Euno's server fits the split Euno's docs publish a tool table covering metadata, lineage, EQL, audit, documentation, catalog and benchmark tools. Euno's server gives an agent governed context about the data; it does not run the query, which is the job of the warehouse's own server. The published facts line up with that statement. No listed tool executes SQL against a warehouse, and per Euno's docs query_to_eql converts a filter without executing it. EQL tools search_resources_eql, count_resources_eql, multi_count_resources_eql and pivot_resources_eql query Euno's metadata graph, not warehouse rows. Euno's site says Euno reads metadata only, and its FAQ says Euno does not query customer data. Per Euno's October 2026 release notes, pivot and multi-count aggregation tools were added through MCP. An engineer wiring an agent would pair a server like this with a query server, and the page for [MCP servers for Euno](/mcp/euno/) holds the record. ## Freshness differs because the two kinds read different things Query servers read live engine state at call time. The Redshift README describes metadata queried live from the clusters, and the community mcp-snowflake-server reads metadata live from Snowflake on each tool call. Metadata servers read what was collected into the catalog. Per Atlan's security page, metadata is retrieved in real time with no intermediate caching, but the catalog contents depend on Atlan's connectors. DataHub's README describes no ingestion cadence, so the server reads whatever DataHub has ingested (none found on 10 Nov 2026). The same finding is recorded for the Collibra README, Alation's README and lineage guide, and Select Star's docs page. An engineer checking whether an answer is current should ask when the catalog last collected from the source, because the MCP server does not state it. ## Pairing the two kinds in one agent An agent often calls a metadata server to find which table holds a metric and who owns it, then calls a query server to read rows. Two servers mean two sets of tool definitions in context, which the [Context Weight Index](/context-weight/) measures, and two identities to check. Databricks documents Genie One as answering natural-language analytics questions grounded in Genie Ontology through one managed MCP, and Cube's runQuery runs Cube SQL against views of its semantic model. The table lists a sample in alphabetical order by what each server describes and whether it executes SQL. | Server | Describes data | Runs or retrieves SQL | | --- | --- | --- | | Atlan MCP (hosted) | Assets, glossary, lineage, tags | query_assets, read-only SQL, up to 100 rows | | BigQuery remote MCP server | Dataset and table metadata | execute_sql_readonly, execute_sql | | Collibra MCP Server | Glossary, lineage, data quality rules | none found on 10 Nov 2026 | | DataHub MCP Server | Search, lineage, owners, tags | get_dataset_queries retrieves query history | | Euno MCP | Governed context, lineage, impact | No listed tool executes SQL | | Monte Carlo MCP Server | Alerts, monitors, asset lineage | none found on 10 Nov 2026 | | Secoda MCP | Search, lineage, glossary | run_sql | | Sifflet MCP Server | Assets, incidents, monitors | none found on 10 Nov 2026 | ## Sources - DataHub MCP README: https://raw.githubusercontent.com/acryldata/mcp-server-datahub/main/README.md (read 10 Nov 2026) - Collibra MCP README: https://github.com/collibra/chip/blob/main/README.md (read 10 Nov 2026) - Select Star MCP product page: https://selectstar.com/product/mcp-for-data (read 10 Nov 2026) - Select Star MCP docs: https://docs.selectstar.com/features/mcp-server (read 10 Nov 2026) - Monte Carlo MCP server docs: https://docs.getmontecarlo.com/docs/mcp-server (read 10 Nov 2026) - Sifflet MCP README: https://github.com/siffletdata/sifflet-mcp/blob/main/README.md (read 10 Nov 2026) - Sifflet MCP server source: https://github.com/siffletdata/sifflet-mcp/blob/1f390ca06d7919214bfe067991539502fdb4b9fc/src/sifflet_mcp/server.py (read 10 Nov 2026) - dbt MCP README: https://raw.githubusercontent.com/dbt-labs/dbt-mcp/main/README.md (read 10 Nov 2026) - AWS Labs Redshift MCP README: https://raw.githubusercontent.com/awslabs/mcp/main/src/redshift-mcp-server/README.md (read 10 Nov 2026) - BigQuery MCP tools reference: https://docs.cloud.google.com/bigquery/docs/reference/mcp (read 10 Nov 2026) - Snowflake-managed MCP docs: https://docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-mcp (read 10 Nov 2026) - Databricks managed MCP page: https://docs.databricks.com/aws/en/generative-ai/mcp/managed-mcp (read 10 Nov 2026) - Atlan MCP tools page: https://docs.atlan.com/product/capabilities/atlan-ai/references/mcp-tools (read 10 Nov 2026) - Atlan MCP security page: https://docs.atlan.com/product/capabilities/atlan-ai/references/mcp-security (read 10 Nov 2026) - Secoda MCP docs: https://docs.secoda.co/features/ai-assistant/secoda-mcp-server.md (read 10 Nov 2026) - Alation authentication guide: https://github.com/Alation/alation-ai-agent-sdk/blob/main/guides/authentication.md (read 10 Nov 2026) - Alation agent SDK README: https://github.com/Alation/alation-ai-agent-sdk/blob/main/README.md (read 10 Nov 2026) - Euno MCP setup docs: https://docs.euno.ai/ai/mcp-setup.md (read 10 Nov 2026) - Euno October 2026 release notes: https://docs.euno.ai/support-resources/release-note/october-2026.md (read 10 Nov 2026) - Euno website and FAQ: https://euno.ai/ (read 10 Nov 2026) - Community mcp-snowflake-server README: https://raw.githubusercontent.com/isaacwasserman/mcp-snowflake-server/main/README.md (read 10 Nov 2026) - Cube MCP server docs: https://docs.cube.dev/docs/integrations/mcp-server.md (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/guides/metadata-servers-versus-query-servers/ --- # Eight facts, all readable from a server's repository or docs in about ten minutes, show what an MCP server for data will do before it touches a warehouse GUIDE, read 10 Nov 2026 **The short answer.** Eight checks cover publisher, version, last release, transport, auth, write tools, tool count and where the tool list is published. Some answers are one line, such as Collibra v0.0.67 or ThoughtSpot 0.5.0, and some are gaps: Sigma publishes no tool names, and Select Star's docs page carries no version or release date. ## Start with the publisher and check the claim against the repository The first line to write down is who publishes the server. For a repository, read the organization that owns it. Fivetran, Collibra, dbt Labs and Sifflet each publish from an organization of their own name. For a hosted server, read which vendor's docs describe it and whether the endpoint is theirs. Look for a support level. Tableau's README carries a badge reading Tableau Supported. Hex's docs describe an official Hex plugin in the Cursor Marketplace. The community mcp-snowflake-server's README says it is a community project and not published by Snowflake. A label such as Labs is a publisher class and not a promise of upkeep, as the [guide on what official means](/guides/) records. Also check for lookalike packages. Fivetran's README says the PyPI names fivetran-mcp and mcp-fivetran belong to unrelated projects, so it tells readers to run its server from the git URL. ## Record the version and the last release as two separate facts A version names what you install. A last-release date says whether anyone is shipping. Record both and the place you read them. For [the dbt MCP Server](/servers/dbt-mcp/), GitHub shows v2.10.1 as latest with a release timestamp of 3 Nov 2026, and PyPI shows the same date. Three gaps are worth writing down as gaps. Hosted servers often publish no version: Select Star's page carries none, and the docs page was last updated seven months before the 14 Jun 2026 archive snapshot that was read. Secoda's repository has no releases, so the date to record is the 26 Jan 2026 commit. ThoughtSpot's releases page shows none, and the npm package's 0.5.0 is dated 6 Jun 2025. A repository's last push date and a package registry's upload date can differ. For the community Snowflake server the PyPI release is 10 Apr 2025 while the GitHub repository was pushed on 7 Oct 2025, so the directory records both and names the source of each. ## Read the transport and the auth route before the config file The transport decides what you run: a local process over stdio, a hosted URL over Streamable HTTP, or both. The [guide on remote versus local servers](/guides/) lists the options. Check whether your client can reach a remote server directly or needs a bridge. Auth decides whose permissions the agent uses. Look for the word OAuth, then for what happens without it. Per Sigma's help page, the API-key route is labeled Beta. Per Monte Carlo's docs, standard API keys do not work and an MCP-specific key is required. Per Alation's README, the remote server does not support dynamic client registration, so a client_id and client_secret must be supplied. Write down which identity the call runs as; the guide on [scoping context](/guides/) covers it. ## List the write tools by name and find the control Open the tool list and search it for create, update, delete, run, deploy, publish, merge, set and put. Then find the page that describes the control. Redshift's README names ACCESS_MODE, DataHub's names TOOLS_IS_MUTATION_ENABLED, Fivetran's names FIVETRAN_SCOPE, and Monte Carlo's names an x-mcp-readonly header. Where the control is missing from the page, write that down too: Secoda's docs list run_sql and describe no read-only mode (none found on 10 Nov 2026), and Sifflet's README describes no per-call prompt for its two incident-status tools (none found on 10 Nov 2026). Check whether the control is enforced by the server or advised to the client; Fivetran's README says write and delete confirmation is advisory only. ## Count the tools and find where the list is published The count tells you roughly how much of the model's context the definitions take, and the [Context Weight Index](/context-weight/) gives the measured size for the servers it covers. Also note where the list is published, because a count is only checkable if the list is. Euno publishes a table of names and descriptions with no parameter schemas. Sigma's page describes capabilities and publishes no tool names. The Databricks Labs README says tools are inferred at startup from a Unity Catalog schema, and Snowflake's managed server defines tools in a specification that you write. Source files can be the list. Sifflet's tool names were read from server.py at commit 1f390ca, ThoughtSpot's from tool-definitions.ts at commit 820bf16f6aa3, and Tableau publishes its annotations in source. Where a docs host fails to resolve, say so: the Select Star page was read from a web.archive.org snapshot, and the snapshot date belongs next to the facts. ## Runtime requirements, licenses and usage limits belong on the same sheet A server can pass every check above and still need something the engineer has not got. Fivetran's README requires Python 3.10 or later and uv. Tableau's package needs Node 22.7.5 or later for npx. Collibra's server is a Go binary, so there is nothing to install beyond the release file. Microsoft's README says use is subject to an EULA and NPX installs require explicit acceptance, and ThoughtSpot's npm package is under an end user license agreement. Usage limits are stated in a few places. Per Airbyte's docs, each organization gets 100,000 direct reads and searches per month at no cost, and other MCP tools do not count toward the limit. Per Collibra's README, natural-language discovery tools use Collibra AI and consume Collibra Units. Google's BigQuery guide says the remote server has no quotas of its own beyond the underlying API quotas. The dbt Labs README adds a warning to read before install: its dbt CLI tools can modify data models and warehouse objects, so proceed only if you trust the client. Collibra's README says its HTTP mode binds to localhost only and that credentials should be kept in environment variables. ## The checklist as a table The table gives the eight checks, where each is usually found, and an example fact from the directory on 10 Nov 2026. The examples are drawn from different servers and are not a comparison of them. The examples in the right-hand column come from different servers on purpose. Each shows what the check looks like when the answer is found, and several sections above show what it looks like when the page is silent. | Check | Where to read it | Example fact | | --- | --- | --- | | Publisher | Repository organization; docs site; support badge | Tableau Supported badge on the Tableau MCP README | | Version | Releases page; package registry; docs stamp | Collibra v0.0.67; Power BI authoring server 1.0.1 | | Last release | Release timestamp; registry upload date | Redshift server 0.1.3 uploaded 9 Nov 2026 | | Transport | README client config; docs configuration table | Sigma configuration table sets transport to HTTP | | Auth | Auth page; environment variables | SYNQ OAuth 2 with PKCE; legacy workspace token | | Write tools | Tool list; read-only flag or header | Redshift ACCESS_MODE; Monte Carlo x-mcp-readonly | | Tool count | Tool table or README list | Atlan's page states 41 tools across nine categories | | Where the list is published | Docs page, source file, or the account specification | Sifflet server.py at commit 1f390ca | ## Sources - Fivetran MCP README: https://raw.githubusercontent.com/fivetran/fivetran-mcp/main/README.md (read 10 Nov 2026) - Collibra chip GitHub releases: https://github.com/collibra/chip/releases (read 10 Nov 2026) - dbt-mcp GitHub releases: https://github.com/dbt-labs/dbt-mcp/releases (read 10 Nov 2026) - Sifflet MCP server source: https://github.com/siffletdata/sifflet-mcp/blob/1f390ca06d7919214bfe067991539502fdb4b9fc/src/sifflet_mcp/server.py (read 10 Nov 2026) - Tableau MCP README: https://raw.githubusercontent.com/tableau/tableau-mcp/main/README.md (read 10 Nov 2026) - Hex MCP server docs: https://learn.hex.tech/docs/api-integrations/mcp-server (read 10 Nov 2026) - Community mcp-snowflake-server README: https://raw.githubusercontent.com/isaacwasserman/mcp-snowflake-server/main/README.md (read 10 Nov 2026) - PyPI mcp-snowflake-server: https://pypi.org/pypi/mcp-snowflake-server/json (read 10 Nov 2026) - Select Star MCP docs (archive snapshot): https://web.archive.org/web/20260614045618/https://docs.selectstar.com/features/mcp-server (read 10 Nov 2026) - Secoda MCP GitHub releases: https://github.com/secoda/secoda-mcp/releases (read 10 Nov 2026) - Secoda MCP latest commit: https://api.github.com/repos/secoda/secoda-mcp/commits?per_page=1 (read 10 Nov 2026) - Secoda MCP docs: https://docs.secoda.co/features/ai-assistant/secoda-mcp-server.md (read 10 Nov 2026) - ThoughtSpot MCP GitHub releases: https://github.com/thoughtspot/mcp-server/releases (read 10 Nov 2026) - npm @thoughtspot/mcp-server: https://registry.npmjs.org/@thoughtspot/mcp-server (read 10 Nov 2026) - ThoughtSpot tool definitions: https://raw.githubusercontent.com/thoughtspot/mcp-server/main/src/servers/tool-definitions.ts (read 10 Nov 2026) - Sigma MCP server help page: https://help.sigmacomputing.com/docs/use-sigma-mcp-server (read 10 Nov 2026) - Monte Carlo MCP server docs: https://docs.getmontecarlo.com/docs/mcp-server (read 10 Nov 2026) - Monte Carlo MCP reference: https://docs.getmontecarlo.com/docs/mcp-server-reference-and-operations (read 10 Nov 2026) - Alation agent SDK README: https://github.com/Alation/alation-ai-agent-sdk/blob/main/README.md (read 10 Nov 2026) - AWS Labs Redshift MCP README: https://raw.githubusercontent.com/awslabs/mcp/main/src/redshift-mcp-server/README.md (read 10 Nov 2026) - PyPI awslabs.redshift-mcp-server: https://pypi.org/pypi/awslabs.redshift-mcp-server/json (read 10 Nov 2026) - DataHub MCP README: https://raw.githubusercontent.com/acryldata/mcp-server-datahub/main/README.md (read 10 Nov 2026) - Sifflet MCP README: https://github.com/siffletdata/sifflet-mcp/blob/main/README.md (read 10 Nov 2026) - Euno MCP setup docs: https://docs.euno.ai/ai/mcp-setup.md (read 10 Nov 2026) - Databricks Labs MCP README: https://raw.githubusercontent.com/databrickslabs/mcp/master/README.md (read 10 Nov 2026) - Snowflake-managed MCP docs: https://docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-mcp (read 10 Nov 2026) - Atlan MCP tools page: https://docs.atlan.com/product/capabilities/atlan-ai/references/mcp-tools (read 10 Nov 2026) - SYNQ MCP docs: https://docs.synq.io/scout/mcp.md (read 10 Nov 2026) - npm @microsoft/powerbi-modeling-mcp: https://registry.npmjs.org/@microsoft/powerbi-modeling-mcp (read 10 Nov 2026) - Airbyte MCP overview: https://docs.airbyte.com/platform/airbyte-mcp.md (read 10 Nov 2026) - Collibra MCP README: https://github.com/collibra/chip/blob/main/README.md (read 10 Nov 2026) - BigQuery remote MCP guide: https://docs.cloud.google.com/bigquery/docs/use-bigquery-mcp (read 10 Nov 2026) - dbt MCP README: https://raw.githubusercontent.com/dbt-labs/dbt-mcp/main/README.md (read 10 Nov 2026) - Power BI Modeling MCP README: https://raw.githubusercontent.com/microsoft/powerbi-modeling-mcp/main/README.md (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/guides/reading-a-server-before-you-install-it/ --- # 10 guides for the engineer who writes down what each server really does GUIDES, read 10 Nov 2026 **The short answer.** Ten guides for an engineer wiring an agent to data tools over MCP: what a server for data should expose, read versus write, scoping context to the person asking, why tool-definition size matters, auth and audit, remote versus local, what official means, and reading a server before installing it, each from the servers' own sources. - [What an MCP server for data should expose](/guides/what-an-mcp-server-for-data-should-expose/): Agents meet five kinds of tools on data servers: query, metadata, lineage, semantic layer and write actions. - [Read versus write on data MCP servers](/guides/read-versus-write/): Redshift prompts for confirmation on each write, DataHub keeps mutation tools off by default, Fivetran defaults its scope to read, and Select Star lists 7 tools that only search, get or traverse. - [Scoping context to the person asking](/guides/scoping-context-to-the-person-asking/): Atlan, Cube, Hex, Omni and Tableau document calls that run as the signed-in user, while Fivetran uses one API key and secret, the Secoda wrapper one API_TOKEN, and Select Star one API token. - [Why tool definition size matters](/guides/why-tool-definition-size-matters/): A client that connects to an MCP server loads each tool's name, description and input schema into the model's context. - [Auth and audit on data MCP servers](/guides/auth-and-audit/): Auth across the directory spans OAuth, personal access tokens, API keys, Basic Auth and cloud credentials. - [Remote versus local MCP servers for data](/guides/remote-versus-local-servers/): Three servers run over stdio only (AWS Labs Redshift, Sifflet, the community mcp-snowflake-server), 15 are hosted-only in the directory's records, and 12 offer both. - [What official means for a data MCP server](/guides/what-official-means/): The directory marks 29 of 30 servers official, meaning the data tool's vendor publishes them; one community server is marked unofficial. - [Three ways to connect an agent to Snowflake](/guides/three-ways-to-connect-an-agent-to-snowflake/): Snowflake documents a managed MCP server created with CREATE MCP SERVER; Snowflake Labs' repository, last released as 1.4.2 on 15 May 2026, is marked deprecated; and the community mcp-snowflake-server, 0.4.0 from 10 Apr 2025, runs over stdio. - [Metadata servers versus query servers](/guides/metadata-servers-versus-query-servers/): Metadata servers such as Atlan, Collibra, DataHub, Monte Carlo and Sifflet describe assets, owners, lineage and incidents. - [Reading a server before you install it](/guides/reading-a-server-before-you-install-it/): Eight checks cover publisher, version, last release, transport, auth, write tools, tool count and where the tool list is published. Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/guides/ --- # Column-level lineage, defined with a sourced example GLOSSARY, read 10 Nov 2026 **The short answer.** Lineage that follows individual columns, showing which source columns feed which output column. An agent uses it to check what a change to one field would touch, not only which tables are linked. Per Alation's guide, its lineage tool traverses upstream or downstream from a starting object and supports column-level lineage, in beta that Alation support must enable. ## Definition Lineage that follows individual columns, showing which source columns feed which output column. An agent uses it to check what a change to one field would touch, not only which tables are linked. ## Example from a real server **Example.** Per Alation's guide, its lineage tool traverses upstream or downstream from a starting object and supports column-level lineage, in beta that Alation support must enable. ## Related terms - [context window](/glossary/context-window/): The limit on how much text a model can read at once, including the conversation, tool definitions and every tool result. - [Cortex Analyst](/glossary/cortex-analyst/): Snowflake's natural-language analytics service. - [write tool](/glossary/write-tool/): A tool that creates, changes or deletes something: a table row, a monitor, a report, a tag. ## Where to read more - [The directory](/) - [All glossary terms](/glossary/) ## Sources - github.com/Alation/alation-ai-agent-sdk/blob/main/guides/tools/lineage.md: https://github.com/Alation/alation-ai-agent-sdk/blob/main/guides/tools/lineage.md (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/glossary/column-level-lineage/ --- # Context window, defined with a sourced example GLOSSARY, read 10 Nov 2026 **The short answer.** The limit on how much text a model can read at once, including the conversation, tool definitions and every tool result. Long results or many tools use up room the model needs for the task. Per Atlan's docs, query_assets runs read-only SQL against a connected warehouse and returns up to 100 rows, which caps the result text reaching the model. ## Definition The limit on how much text a model can read at once, including the conversation, tool definitions and every tool result. Long results or many tools use up room the model needs for the task. ## Example from a real server **Example.** Per Atlan's docs, query_assets runs read-only SQL against a connected warehouse and returns up to 100 rows, which caps the result text reaching the model. ## Related terms - [Cortex Analyst](/glossary/cortex-analyst/): Snowflake's natural-language analytics service. - [data observability incident](/glossary/data-observability-incident/): A recorded problem found by monitoring, such as late data, a failed test or a volume change. - [column-level lineage](/glossary/column-level-lineage/): Lineage that follows individual columns, showing which source columns feed which output column. ## Where to read more - [The directory](/) - [All glossary terms](/glossary/) ## Sources - docs.atlan.com/product/capabilities/atlan-ai/references/mcp-tools: https://docs.atlan.com/product/capabilities/atlan-ai/references/mcp-tools (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/glossary/context-window/ --- # Cortex Analyst, defined with a sourced example GLOSSARY, read 10 Nov 2026 **The short answer.** Snowflake's natural-language analytics service. It answers questions by generating SQL against a semantic view, which holds business definitions. A Snowflake MCP server can offer it to agents as a tool. Per Snowflake's docs, Cortex Analyst tools use semantic views, and the managed MCP server accepts semantic views only, not semantic models. ## Definition Snowflake's natural-language analytics service. It answers questions by generating SQL against a semantic view, which holds business definitions. A Snowflake MCP server can offer it to agents as a tool. ## Example from a real server **Example.** Per Snowflake's docs, Cortex Analyst tools use semantic views, and the managed MCP server accepts semantic views only, not semantic models. ## Related terms - [data observability incident](/glossary/data-observability-incident/): A recorded problem found by monitoring, such as late data, a failed test or a volume change. - [Discovery API](/glossary/discovery-api/): The dbt platform API that returns metadata about a project's models, sources, exposures, tests and runs. - [context window](/glossary/context-window/): The limit on how much text a model can read at once, including the conversation, tool definitions and every tool result. ## Where to read more - [The directory](/) - [All glossary terms](/glossary/) ## Sources - docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-mcp: https://docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-mcp (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/glossary/cortex-analyst/ --- # Data observability incident, defined with a sourced example GLOSSARY, read 10 Nov 2026 **The short answer.** A recorded problem found by monitoring, such as late data, a failed test or a volume change. An agent can open it, read what it affects and sometimes change its status. Per the source of Sifflet's MCP server, two tools change state: close_incident_by_id and open_incident_by_id patch an incident's status. ## Definition A recorded problem found by monitoring, such as late data, a failed test or a volume change. An agent can open it, read what it affects and sometimes change its status. ## Example from a real server **Example.** Per the source of Sifflet's MCP server, two tools change state: close_incident_by_id and open_incident_by_id patch an incident's status. ## Related terms - [Discovery API](/glossary/discovery-api/): The dbt platform API that returns metadata about a project's models, sources, exposures, tests and runs. - [dynamic client registration](/glossary/dynamic-client-registration/): A step where an MCP client registers itself with an authorization server automatically, so nobody has to create a client ID by hand. - [Cortex Analyst](/glossary/cortex-analyst/): Snowflake's natural-language analytics service. ## Where to read more - [The directory](/) - [All glossary terms](/glossary/) ## Sources - github.com/siffletdata/sifflet-mcp/blob/1f390ca06d7919214bfe067991539502fdb4b9fc/src/sifflet_mcp/server.py: https://github.com/siffletdata/sifflet-mcp/blob/1f390ca06d7919214bfe067991539502fdb4b9fc/src/sifflet_mcp/server.py (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/glossary/data-observability-incident/ --- # Discovery API, defined with a sourced example GLOSSARY, read 10 Nov 2026 **The short answer.** The dbt platform API that returns metadata about a project's models, sources, exposures, tests and runs. It describes the project and its last runs, not the rows in the warehouse. Per dbt Labs' README, the server's discovery tools cover models, sources, exposures and freshness, and get_model_health returns run status, test results and upstream source freshness. ## Definition The dbt platform API that returns metadata about a project's models, sources, exposures, tests and runs. It describes the project and its last runs, not the rows in the warehouse. ## Example from a real server **Example.** Per dbt Labs' README, the server's discovery tools cover models, sources, exposures and freshness, and get_model_health returns run status, test results and upstream source freshness. ## Related terms - [dynamic client registration](/glossary/dynamic-client-registration/): A step where an MCP client registers itself with an authorization server automatically, so nobody has to create a client ID by hand. - [Genie space](/glossary/genie-space/): A Databricks workspace for asking questions of chosen tables in natural language, set up with curated instructions. - [data observability incident](/glossary/data-observability-incident/): A recorded problem found by monitoring, such as late data, a failed test or a volume change. ## Where to read more - [The directory](/) - [All glossary terms](/glossary/) ## Sources - raw.githubusercontent.com/dbt-labs/dbt-mcp/main/README.md: https://raw.githubusercontent.com/dbt-labs/dbt-mcp/main/README.md (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/glossary/discovery-api/ --- # Dynamic client registration, defined with a sourced example GLOSSARY, read 10 Nov 2026 **The short answer.** A step where an MCP client registers itself with an authorization server automatically, so nobody has to create a client ID by hand. Servers that skip it make each team supply a client ID and secret. Per Alation's README, its remote MCP server does not support dynamic client registration, so clients must supply a client_id and client_secret. ## Definition A step where an MCP client registers itself with an authorization server automatically, so nobody has to create a client ID by hand. Servers that skip it make each team supply a client ID and secret. ## Example from a real server **Example.** Per Alation's README, its remote MCP server does not support dynamic client registration, so clients must supply a client_id and client_secret. ## Related terms - [Genie space](/glossary/genie-space/): A Databricks workspace for asking questions of chosen tables in natural language, set up with curated instructions. - [human confirmation](/glossary/human-confirmation/): A step where a person approves a tool call before it runs. - [Discovery API](/glossary/discovery-api/): The dbt platform API that returns metadata about a project's models, sources, exposures, tests and runs. ## Where to read more - [The directory](/) - [All glossary terms](/glossary/) ## Sources - github.com/Alation/alation-ai-agent-sdk/blob/main/README.md: https://github.com/Alation/alation-ai-agent-sdk/blob/main/README.md (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/glossary/dynamic-client-registration/ --- # Genie space, defined with a sourced example GLOSSARY, read 10 Nov 2026 **The short answer.** A Databricks workspace for asking questions of chosen tables in natural language, set up with curated instructions. An MCP server can expose a space as a tool that takes a question and returns an answer. Per Databricks' docs, Genie Agent MCP calls bypass Unity Gateway and rely on the permissions of the Genie Agent and its data. ## Definition A Databricks workspace for asking questions of chosen tables in natural language, set up with curated instructions. An MCP server can expose a space as a tool that takes a question and returns an answer. ## Example from a real server **Example.** Per Databricks' docs, Genie Agent MCP calls bypass Unity Gateway and rely on the permissions of the Genie Agent and its data. ## Related terms - [human confirmation](/glossary/human-confirmation/): A step where a person approves a tool call before it runs. - [impact analysis](/glossary/impact-analysis/): The question of what breaks downstream if an asset changes or fails. - [dynamic client registration](/glossary/dynamic-client-registration/): A step where an MCP client registers itself with an authorization server automatically, so nobody has to create a client ID by hand. ## Where to read more - [The directory](/) - [All glossary terms](/glossary/) ## Sources - docs.databricks.com/aws/en/generative-ai/mcp/managed-mcp: https://docs.databricks.com/aws/en/generative-ai/mcp/managed-mcp (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/glossary/genie-space/ --- # Human confirmation, defined with a sourced example GLOSSARY, read 10 Nov 2026 **The short answer.** A step where a person approves a tool call before it runs. The MCP specification keeps a human in the loop with the ability to deny invocations. Some servers ask through the protocol. Per AWS Labs' README, in read-write mode the Redshift server asks for confirmation on each write through MCP elicitation, unless UNSAFE_SKIP_WRITE_CONFIRMATION is set. ## Definition A step where a person approves a tool call before it runs. The MCP specification keeps a human in the loop with the ability to deny invocations. Some servers ask through the protocol. ## Example from a real server **Example.** Per AWS Labs' README, in read-write mode the Redshift server asks for confirmation on each write through MCP elicitation, unless UNSAFE_SKIP_WRITE_CONFIRMATION is set. ## Related terms - [impact analysis](/glossary/impact-analysis/): The question of what breaks downstream if an asset changes or fails. - [input schema](/glossary/input-schema/): A JSON Schema that states which arguments a tool accepts, their types and which are required or limited to set values. - [Genie space](/glossary/genie-space/): A Databricks workspace for asking questions of chosen tables in natural language, set up with curated instructions. ## Where to read more - [The directory](/) - [All glossary terms](/glossary/) ## Sources - raw.githubusercontent.com/awslabs/mcp/main/src/redshift-mcp-server/README.md: https://raw.githubusercontent.com/awslabs/mcp/main/src/redshift-mcp-server/README.md (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/glossary/human-confirmation/ --- # Impact analysis, defined with a sourced example GLOSSARY, read 10 Nov 2026 **The short answer.** The question of what breaks downstream if an asset changes or fails. An agent answers it by walking dependencies from the asset to the dashboards, models and reports that read from it. Per Sifflet's README, the server can start from an incident and trace the downstream assets affected, and its get_downstream_assets_of_asset tool returns them by URN. ## Definition The question of what breaks downstream if an asset changes or fails. An agent answers it by walking dependencies from the asset to the dashboards, models and reports that read from it. ## Example from a real server **Example.** Per Sifflet's README, the server can start from an incident and trace the downstream assets affected, and its get_downstream_assets_of_asset tool returns them by URN. ## Related terms - [input schema](/glossary/input-schema/): A JSON Schema that states which arguments a tool accepts, their types and which are required or limited to set values. - [local MCP server](/glossary/local-mcp-server/): An MCP server that runs on the engineer's own machine or infrastructure, usually started by the client over stdio. - [human confirmation](/glossary/human-confirmation/): A step where a person approves a tool call before it runs. ## Where to read more - [The directory](/) - [All glossary terms](/glossary/) ## Sources - github.com/siffletdata/sifflet-mcp/blob/main/README.md: https://github.com/siffletdata/sifflet-mcp/blob/main/README.md (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/glossary/impact-analysis/ --- # Input schema, defined with a sourced example GLOSSARY, read 10 Nov 2026 **The short answer.** A JSON Schema that states which arguments a tool accepts, their types and which are required or limited to set values. A tighter schema gives the model less room to send a malformed or unintended call. Per the source of Sifflet's MCP server, the search_asset tool filters by asset type, with values TABLE_AND_VIEW, PIPELINE, DASHBOARD or ML_MODEL. ## Definition A JSON Schema that states which arguments a tool accepts, their types and which are required or limited to set values. A tighter schema gives the model less room to send a malformed or unintended call. ## Example from a real server **Example.** Per the source of Sifflet's MCP server, the search_asset tool filters by asset type, with values TABLE_AND_VIEW, PIPELINE, DASHBOARD or ML_MODEL. ## Related terms - [local MCP server](/glossary/local-mcp-server/): An MCP server that runs on the engineer's own machine or infrastructure, usually started by the client over stdio. - [MCP client](/glossary/mcp-client/): The part of an AI application that connects to MCP servers, lists their tools and sends the calls the model decides to make. - [impact analysis](/glossary/impact-analysis/): The question of what breaks downstream if an asset changes or fails. ## Where to read more - [The directory](/) - [All glossary terms](/glossary/) ## Sources - github.com/siffletdata/sifflet-mcp/blob/1f390ca06d7919214bfe067991539502fdb4b9fc/src/sifflet_mcp/server.py: https://github.com/siffletdata/sifflet-mcp/blob/1f390ca06d7919214bfe067991539502fdb4b9fc/src/sifflet_mcp/server.py (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/glossary/input-schema/ --- # Local MCP server, defined with a sourced example GLOSSARY, read 10 Nov 2026 **The short answer.** An MCP server that runs on the engineer's own machine or infrastructure, usually started by the client over stdio. The team controls the version and credentials, and must also install and update it. Per Sifflet's README, setup for Cursor and Claude Desktop runs the server locally through uvx sifflet-mcp@latest, with a Sifflet API token in environment variables. ## Definition An MCP server that runs on the engineer's own machine or infrastructure, usually started by the client over stdio. The team controls the version and credentials, and must also install and update it. ## Example from a real server **Example.** Per Sifflet's README, setup for Cursor and Claude Desktop runs the server locally through uvx sifflet-mcp@latest, with a Sifflet API token in environment variables. ## Related terms - [MCP client](/glossary/mcp-client/): The part of an AI application that connects to MCP servers, lists their tools and sends the calls the model decides to make. - [MCP server](/glossary/mcp-server/): A program that exposes tools, resources or prompts to an AI application over the Model Context Protocol. - [input schema](/glossary/input-schema/): A JSON Schema that states which arguments a tool accepts, their types and which are required or limited to set values. ## Where to read more - [The directory](/) - [All glossary terms](/glossary/) ## Sources - github.com/siffletdata/sifflet-mcp/blob/main/README.md: https://github.com/siffletdata/sifflet-mcp/blob/main/README.md (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/glossary/local-mcp-server/ --- # MCP client, defined with a sourced example GLOSSARY, read 10 Nov 2026 **The short answer.** The part of an AI application that connects to MCP servers, lists their tools and sends the calls the model decides to make. Chat apps and code editors each embed one. Per Hex's docs, the Hex MCP server works with Claude Desktop, Claude Code, Cursor, ChatGPT, Codex and Glean. ## Definition The part of an AI application that connects to MCP servers, lists their tools and sends the calls the model decides to make. Chat apps and code editors each embed one. ## Example from a real server **Example.** Per Hex's docs, the Hex MCP server works with Claude Desktop, Claude Code, Cursor, ChatGPT, Codex and Glean. ## Related terms - [MCP server](/glossary/mcp-server/): A program that exposes tools, resources or prompts to an AI application over the Model Context Protocol. - [mcp-remote bridge](/glossary/mcp-remote-bridge/): An open-source Node.js proxy that lets a client which only speaks stdio reach a remote MCP server. - [local MCP server](/glossary/local-mcp-server/): An MCP server that runs on the engineer's own machine or infrastructure, usually started by the client over stdio. ## Where to read more - [The directory](/) - [All glossary terms](/glossary/) ## Sources - learn.hex.tech/docs/api-integrations/mcp-server: https://learn.hex.tech/docs/api-integrations/mcp-server (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/glossary/mcp-client/ --- # MCP server, defined with a sourced example GLOSSARY, read 10 Nov 2026 **The short answer.** A program that exposes tools, resources or prompts to an AI application over the Model Context Protocol. For data work, it sits in front of a warehouse, a dbt project, a BI tool or a catalog. Per dbt Labs' README, the dbt MCP server gives AI agents context of a project in dbt Core, dbt Fusion and dbt Platform. ## Definition A program that exposes tools, resources or prompts to an AI application over the Model Context Protocol. For data work, it sits in front of a warehouse, a dbt project, a BI tool or a catalog. ## Example from a real server **Example.** Per dbt Labs' README, the dbt MCP server gives AI agents context of a project in dbt Core, dbt Fusion and dbt Platform. ## Related terms - [mcp-remote bridge](/glossary/mcp-remote-bridge/): An open-source Node.js proxy that lets a client which only speaks stdio reach a remote MCP server. - [metadata graph](/glossary/metadata-graph/): A connected store of data assets and their relationships: tables, columns, dashboards, owners and the links between them. - [MCP client](/glossary/mcp-client/): The part of an AI application that connects to MCP servers, lists their tools and sends the calls the model decides to make. ## Where to read more - [The directory](/) - [All glossary terms](/glossary/) ## Sources - raw.githubusercontent.com/dbt-labs/dbt-mcp/main/README.md: https://raw.githubusercontent.com/dbt-labs/dbt-mcp/main/README.md (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/glossary/mcp-server/ --- # Mcp-remote bridge, defined with a sourced example GLOSSARY, read 10 Nov 2026 **The short answer.** An open-source Node.js proxy that lets a client which only speaks stdio reach a remote MCP server. It runs locally, forwards calls to the remote URL, and can pass an API token as a header. Per Euno's docs, stdio use goes through the third-party open-source mcp-remote bridge, not a package that Euno publishes. ## Definition An open-source Node.js proxy that lets a client which only speaks stdio reach a remote MCP server. It runs locally, forwards calls to the remote URL, and can pass an API token as a header. ## Example from a real server **Example.** Per Euno's docs, stdio use goes through the third-party open-source mcp-remote bridge, not a package that Euno publishes. ## Related terms - [metadata graph](/glossary/metadata-graph/): A connected store of data assets and their relationships: tables, columns, dashboards, owners and the links between them. - [monitor as code](/glossary/monitor-as-code/): Defining a data monitor as configuration kept in a repository instead of clicking it together in a UI. - [MCP server](/glossary/mcp-server/): A program that exposes tools, resources or prompts to an AI application over the Model Context Protocol. ## Where to read more - [The directory](/) - [All glossary terms](/glossary/) ## Sources - docs.euno.ai/ai/mcp-setup/mcp-bridge-setup.md: https://docs.euno.ai/ai/mcp-setup/mcp-bridge-setup.md (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/glossary/mcp-remote-bridge/ --- # Metadata graph, defined with a sourced example GLOSSARY, read 10 Nov 2026 **The short answer.** A connected store of data assets and their relationships: tables, columns, dashboards, owners and the links between them. An agent searches or walks it to find what exists and how it connects. Per Euno's docs, its MCP server can search and aggregate the metadata graph with EQL through search, count, multi-count and pivot tools. ## Definition A connected store of data assets and their relationships: tables, columns, dashboards, owners and the links between them. An agent searches or walks it to find what exists and how it connects. ## Example from a real server **Example.** Per Euno's docs, its MCP server can search and aggregate the metadata graph with EQL through search, count, multi-count and pivot tools. ## Related terms - [monitor as code](/glossary/monitor-as-code/): Defining a data monitor as configuration kept in a repository instead of clicking it together in a UI. - [OAuth 2.1 authorization for MCP](/glossary/oauth-2-1-authorization-for-mcp/): The sign-in flow the MCP specification defines for HTTP servers. - [mcp-remote bridge](/glossary/mcp-remote-bridge/): An open-source Node.js proxy that lets a client which only speaks stdio reach a remote MCP server. ## Where to read more - [The directory](/) - [All glossary terms](/glossary/) ## Sources - docs.euno.ai/ai/mcp-setup.md: https://docs.euno.ai/ai/mcp-setup.md (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/glossary/metadata-graph/ --- # Monitor as code, defined with a sourced example GLOSSARY, read 10 Nov 2026 **The short answer.** Defining a data monitor as configuration kept in a repository instead of clicking it together in a UI. An agent can draft the configuration from a description, and a person reviews it before it ships. Per Sifflet's README, the get_monitor_code_by_description tool returns a monitor configuration for a plain-language description and needs the Editor role on the domain. ## Definition Defining a data monitor as configuration kept in a repository instead of clicking it together in a UI. An agent can draft the configuration from a description, and a person reviews it before it ships. ## Example from a real server **Example.** Per Sifflet's README, the get_monitor_code_by_description tool returns a monitor configuration for a plain-language description and needs the Editor role on the domain. ## Related terms - [OAuth 2.1 authorization for MCP](/glossary/oauth-2-1-authorization-for-mcp/): The sign-in flow the MCP specification defines for HTTP servers. - [persona scoping](/glossary/persona-scoping/): Limiting what an agent can see or do by the persona of the signed-in person, such as admin, member or guest. - [metadata graph](/glossary/metadata-graph/): A connected store of data assets and their relationships: tables, columns, dashboards, owners and the links between them. ## Where to read more - [The directory](/) - [All glossary terms](/glossary/) ## Sources - github.com/siffletdata/sifflet-mcp/blob/main/README.md: https://github.com/siffletdata/sifflet-mcp/blob/main/README.md (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/glossary/monitor-as-code/ --- # OAuth 2.1 authorization for MCP, defined with a sourced example GLOSSARY, read 10 Nov 2026 **The short answer.** The sign-in flow the MCP specification defines for HTTP servers. A user approves access in a browser, and the client holds a token scoped to that user. Calls then run with that person's permissions, not a shared secret. Per Monte Carlo's docs, its hosted MCP server supports OAuth 2.1 with Dynamic Client Registration, and requires an Editor role or above. ## Definition The sign-in flow the MCP specification defines for HTTP servers. A user approves access in a browser, and the client holds a token scoped to that user. Calls then run with that person's permissions, not a shared secret. ## Example from a real server **Example.** Per Monte Carlo's docs, its hosted MCP server supports OAuth 2.1 with Dynamic Client Registration, and requires an Editor role or above. ## Related terms - [persona scoping](/glossary/persona-scoping/): Limiting what an agent can see or do by the persona of the signed-in person, such as admin, member or guest. - [personal access token](/glossary/personal-access-token/): A long-lived secret a person creates in a tool to let a script or agent act as them. - [monitor as code](/glossary/monitor-as-code/): Defining a data monitor as configuration kept in a repository instead of clicking it together in a UI. ## Where to read more - [The directory](/) - [All glossary terms](/glossary/) ## Sources - docs.getmontecarlo.com/docs/mcp-server: https://docs.getmontecarlo.com/docs/mcp-server (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/glossary/oauth-2-1-authorization-for-mcp/ --- # Persona scoping, defined with a sourced example GLOSSARY, read 10 Nov 2026 **The short answer.** Limiting what an agent can see or do by the persona of the signed-in person, such as admin, member or guest. The server narrows results to that role, so the agent never exceeds what the login allows. Per Atlan's docs, guests and members are scoped to their persona, and MCP narrows but never widens what a user's login permits. ## Definition Limiting what an agent can see or do by the persona of the signed-in person, such as admin, member or guest. The server narrows results to that role, so the agent never exceeds what the login allows. ## Example from a real server **Example.** Per Atlan's docs, guests and members are scoped to their persona, and MCP narrows but never widens what a user's login permits. ## Related terms - [personal access token](/glossary/personal-access-token/): A long-lived secret a person creates in a tool to let a script or agent act as them. - [read-only tool](/glossary/read-only-tool/): A tool that changes nothing in the system it talks to. - [OAuth 2.1 authorization for MCP](/glossary/oauth-2-1-authorization-for-mcp/): The sign-in flow the MCP specification defines for HTTP servers. ## Where to read more - [The directory](/) - [All glossary terms](/glossary/) ## Sources - docs.atlan.com/product/capabilities/atlan-ai/references/mcp-security: https://docs.atlan.com/product/capabilities/atlan-ai/references/mcp-security (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/glossary/persona-scoping/ --- # Personal access token, defined with a sourced example GLOSSARY, read 10 Nov 2026 **The short answer.** A long-lived secret a person creates in a tool to let a script or agent act as them. Calls carry that person's role and access, so the token must be stored like a password. Per DataHub's README, in the Docker HTTP deployment each MCP client supplies its own DataHub token, so permissions and audit identity stay per user. ## Definition A long-lived secret a person creates in a tool to let a script or agent act as them. Calls carry that person's role and access, so the token must be stored like a password. ## Example from a real server **Example.** Per DataHub's README, in the Docker HTTP deployment each MCP client supplies its own DataHub token, so permissions and audit identity stay per user. ## Related terms - [read-only tool](/glossary/read-only-tool/): A tool that changes nothing in the system it talks to. - [remote MCP server](/glossary/remote-mcp-server/): An MCP server the vendor hosts and the client reaches by URL. - [persona scoping](/glossary/persona-scoping/): Limiting what an agent can see or do by the persona of the signed-in person, such as admin, member or guest. ## Where to read more - [The directory](/) - [All glossary terms](/glossary/) ## Sources - raw.githubusercontent.com/acryldata/mcp-server-datahub/main/README.md: https://raw.githubusercontent.com/acryldata/mcp-server-datahub/main/README.md (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/glossary/personal-access-token/ --- # Read-only tool, defined with a sourced example GLOSSARY, read 10 Nov 2026 **The short answer.** A tool that changes nothing in the system it talks to. Read-only tools can run without a person approving each call. Some servers can hide every other tool, so the client never sees a way to write. Per Airbyte's docs, read-only mode (X-MCP-Readonly-Mode header or ?read-only=true) shows only read-only tools and hides the run and cancel sync tools. ## Definition A tool that changes nothing in the system it talks to. Read-only tools can run without a person approving each call. Some servers can hide every other tool, so the client never sees a way to write. ## Example from a real server **Example.** Per Airbyte's docs, read-only mode (X-MCP-Readonly-Mode header or ?read-only=true) shows only read-only tools and hides the run and cancel sync tools. ## Related terms - [remote MCP server](/glossary/remote-mcp-server/): An MCP server the vendor hosts and the client reaches by URL. - [resource](/glossary/resource/): Read-only context an MCP server exposes by address, such as a file or a database schema. - [personal access token](/glossary/personal-access-token/): A long-lived secret a person creates in a tool to let a script or agent act as them. ## Where to read more - [The directory](/) - [All glossary terms](/glossary/) ## Sources - docs.airbyte.com/platform/airbyte-mcp/advanced-configuration.md: https://docs.airbyte.com/platform/airbyte-mcp/advanced-configuration.md (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/glossary/read-only-tool/ --- # Remote MCP server, defined with a sourced example GLOSSARY, read 10 Nov 2026 **The short answer.** An MCP server the vendor hosts and the client reaches by URL. Nothing is installed, the vendor updates it, and sign-in usually uses OAuth. The version a team runs is whatever the vendor currently serves. Per Sigma's docs, the Sigma MCP server is a hosted endpoint whose URL is copied from Profile > Integrations; no versioned release is published. ## Definition An MCP server the vendor hosts and the client reaches by URL. Nothing is installed, the vendor updates it, and sign-in usually uses OAuth. The version a team runs is whatever the vendor currently serves. ## Example from a real server **Example.** Per Sigma's docs, the Sigma MCP server is a hosted endpoint whose URL is copied from Profile > Integrations; no versioned release is published. ## Related terms - [resource](/glossary/resource/): Read-only context an MCP server exposes by address, such as a file or a database schema. - [row-level security](/glossary/row-level-security/): A rule that filters which rows a person may read in a table or model, often by role or department. - [read-only tool](/glossary/read-only-tool/): A tool that changes nothing in the system it talks to. ## Where to read more - [The directory](/) - [All glossary terms](/glossary/) ## Sources - help.sigmacomputing.com/docs/use-sigma-mcp-server: https://help.sigmacomputing.com/docs/use-sigma-mcp-server (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/glossary/remote-mcp-server/ --- # Resource, defined with a sourced example GLOSSARY, read 10 Nov 2026 **The short answer.** Read-only context an MCP server exposes by address, such as a file or a database schema. The host application decides when to load it, unlike a tool, which the model calls on its own. Per the MCP specification, resources provide context such as files, database schemas or application-specific information, each identified by a URI. ## Definition Read-only context an MCP server exposes by address, such as a file or a database schema. The host application decides when to load it, unlike a tool, which the model calls on its own. ## Example from a real server **Example.** Per the MCP specification, resources provide context such as files, database schemas or application-specific information, each identified by a URI. ## Related terms - [row-level security](/glossary/row-level-security/): A rule that filters which rows a person may read in a table or model, often by role or department. - [semantic layer query](/glossary/semantic-layer-query/): A query written against defined metrics and dimensions instead of raw tables. - [remote MCP server](/glossary/remote-mcp-server/): An MCP server the vendor hosts and the client reaches by URL. ## Where to read more - [The directory](/) - [All glossary terms](/glossary/) ## Sources - modelcontextprotocol.io/specification/2025-11-25/server/resources: https://modelcontextprotocol.io/specification/2025-11-25/server/resources (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/glossary/resource/ --- # Row-level security, defined with a sourced example GLOSSARY, read 10 Nov 2026 **The short answer.** A rule that filters which rows a person may read in a table or model, often by role or department. Two users can ask an agent the same question and get different rows back. Per Microsoft's README, the Power BI modeling MCP server has tools to manage security roles with row-level security, alongside perspectives, cultures and object translations. ## Definition A rule that filters which rows a person may read in a table or model, often by role or department. Two users can ask an agent the same question and get different rows back. ## Example from a real server **Example.** Per Microsoft's README, the Power BI modeling MCP server has tools to manage security roles with row-level security, alongside perspectives, cultures and object translations. ## Related terms - [semantic layer query](/glossary/semantic-layer-query/): A query written against defined metrics and dimensions instead of raw tables. - [service account](/glossary/service-account/): A non-human identity with its own role, used by unattended agents and jobs. - [resource](/glossary/resource/): Read-only context an MCP server exposes by address, such as a file or a database schema. ## Where to read more - [The directory](/) - [All glossary terms](/glossary/) ## Sources - raw.githubusercontent.com/microsoft/powerbi-modeling-mcp/main/README.md: https://raw.githubusercontent.com/microsoft/powerbi-modeling-mcp/main/README.md (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/glossary/row-level-security/ --- # Semantic layer query, defined with a sourced example GLOSSARY, read 10 Nov 2026 **The short answer.** A query written against defined metrics and dimensions instead of raw tables. The semantic layer turns it into SQL, so every agent that asks for revenue gets the same definition of revenue. Per Cube's docs, its runQuery tool runs Cube SQL against views whose names and members come from the searchDataModel tool. ## Definition A query written against defined metrics and dimensions instead of raw tables. The semantic layer turns it into SQL, so every agent that asks for revenue gets the same definition of revenue. ## Example from a real server **Example.** Per Cube's docs, its runQuery tool runs Cube SQL against views whose names and members come from the searchDataModel tool. ## Related terms - [service account](/glossary/service-account/): A non-human identity with its own role, used by unattended agents and jobs. - [stdio transport](/glossary/stdio-transport/): A way to run an MCP server as a local subprocess. - [row-level security](/glossary/row-level-security/): A rule that filters which rows a person may read in a table or model, often by role or department. ## Where to read more - [The directory](/) - [All glossary terms](/glossary/) ## Sources - docs.cube.dev/docs/integrations/mcp-server.md: https://docs.cube.dev/docs/integrations/mcp-server.md (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/glossary/semantic-layer-query/ --- # Service account, defined with a sourced example GLOSSARY, read 10 Nov 2026 **The short answer.** A non-human identity with its own role, used by unattended agents and jobs. Calls are logged under the account, not a person, and the role caps what the agent can reach. Per OpenMetadata's docs, unattended agents should use a Bot service account with a role-based access policy, kept separate from personal logins and auditable. ## Definition A non-human identity with its own role, used by unattended agents and jobs. Calls are logged under the account, not a person, and the role caps what the agent can reach. ## Example from a real server **Example.** Per OpenMetadata's docs, unattended agents should use a Bot service account with a role-based access policy, kept separate from personal logins and auditable. ## Related terms - [stdio transport](/glossary/stdio-transport/): A way to run an MCP server as a local subprocess. - [Streamable HTTP](/glossary/streamable-http/): The MCP transport for servers reached over a network. - [semantic layer query](/glossary/semantic-layer-query/): A query written against defined metrics and dimensions instead of raw tables. ## Where to read more - [The directory](/) - [All glossary terms](/glossary/) ## Sources - docs.open-metadata.org/v2.0.x/how-to-guides/mcp.md: https://docs.open-metadata.org/v2.0.x/how-to-guides/mcp.md (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/glossary/service-account/ --- # Stdio transport, defined with a sourced example GLOSSARY, read 10 Nov 2026 **The short answer.** A way to run an MCP server as a local subprocess. The client starts the process, writes requests to its standard input and reads replies from its standard output. Credentials come from the local environment. Per AWS Labs' README client configuration, the Amazon Redshift MCP server runs over stdio and uses AWS credentials from a profile or environment variables. ## Definition A way to run an MCP server as a local subprocess. The client starts the process, writes requests to its standard input and reads replies from its standard output. Credentials come from the local environment. ## Example from a real server **Example.** Per AWS Labs' README client configuration, the Amazon Redshift MCP server runs over stdio and uses AWS credentials from a profile or environment variables. ## Related terms - [Streamable HTTP](/glossary/streamable-http/): The MCP transport for servers reached over a network. - [text-to-SQL](/glossary/text-to-sql/): A tool that turns a plain-language question into a SQL statement, usually using what it knows about the schema or project. - [service account](/glossary/service-account/): A non-human identity with its own role, used by unattended agents and jobs. ## Where to read more - [The directory](/) - [All glossary terms](/glossary/) ## Sources - raw.githubusercontent.com/awslabs/mcp/main/src/redshift-mcp-server/README.md: https://raw.githubusercontent.com/awslabs/mcp/main/src/redshift-mcp-server/README.md (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/glossary/stdio-transport/ --- # Streamable HTTP, defined with a sourced example GLOSSARY, read 10 Nov 2026 **The short answer.** The MCP transport for servers reached over a network. The client sends HTTP requests to one endpoint, and the server can answer many clients at once. Per Google Cloud's docs, the BigQuery remote MCP server sits at the global endpoint bigquery.googleapis.com/mcp and uses the Streamable HTTP transport. ## Definition The MCP transport for servers reached over a network. The client sends HTTP requests to one endpoint, and the server can answer many clients at once. It replaced the older HTTP plus SSE transport from protocol version 5 Nov 2024. ## Example from a real server **Example.** Per Google Cloud's docs, the BigQuery remote MCP server sits at the global endpoint bigquery.googleapis.com/mcp and uses the Streamable HTTP transport. ## Related terms - [text-to-SQL](/glossary/text-to-sql/): A tool that turns a plain-language question into a SQL statement, usually using what it knows about the schema or project. - [token](/glossary/token/): The unit in which models count text, roughly a word fragment. - [stdio transport](/glossary/stdio-transport/): A way to run an MCP server as a local subprocess. ## Where to read more - [The directory](/) - [All glossary terms](/glossary/) ## Sources - docs.cloud.google.com/bigquery/docs/use-bigquery-mcp: https://docs.cloud.google.com/bigquery/docs/use-bigquery-mcp (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/glossary/streamable-http/ --- # Text-to-SQL, defined with a sourced example GLOSSARY, read 10 Nov 2026 **The short answer.** A tool that turns a plain-language question into a SQL statement, usually using what it knows about the schema or project. The SQL still has to be checked and run, and who runs it varies by server. Per dbt Labs' README, its text_to_sql tool generates SQL from natural language using project context. ## Definition A tool that turns a plain-language question into a SQL statement, usually using what it knows about the schema or project. The SQL still has to be checked and run, and who runs it varies by server. ## Example from a real server **Example.** Per dbt Labs' README, its text_to_sql tool generates SQL from natural language using project context. ## Related terms - [token](/glossary/token/): The unit in which models count text, roughly a word fragment. - [tool](/glossary/tool/): A named function an MCP server offers to the model. - [Streamable HTTP](/glossary/streamable-http/): The MCP transport for servers reached over a network. ## Where to read more - [The directory](/) - [All glossary terms](/glossary/) ## Sources - raw.githubusercontent.com/dbt-labs/dbt-mcp/main/README.md: https://raw.githubusercontent.com/dbt-labs/dbt-mcp/main/README.md (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/glossary/text-to-sql/ --- # Token, defined with a sourced example GLOSSARY, read 10 Nov 2026 **The short answer.** The unit in which models count text, roughly a word fragment. Context windows, limits and prices are stated in tokens. Tool definitions are text, so published tools cost a measurable number of tokens before any call. Per Euno's docs, its MCP server publishes names and descriptions for 56 tools, and each definition is text the model reads as tokens. ## Definition The unit in which models count text, roughly a word fragment. Context windows, limits and prices are stated in tokens. Tool definitions are text, so published tools cost a measurable number of tokens before any call. ## Example from a real server **Example.** Per Euno's docs, its MCP server publishes names and descriptions for 56 tools, and each definition is text the model reads as tokens. ## Related terms - [tool](/glossary/tool/): A named function an MCP server offers to the model. - [tool annotations](/glossary/tool-annotations/): Optional hints a server attaches to a tool, such as readOnlyHint, destructiveHint, idempotentHint and openWorldHint. - [text-to-SQL](/glossary/text-to-sql/): A tool that turns a plain-language question into a SQL statement, usually using what it knows about the schema or project. ## Where to read more - [The directory](/) - [All glossary terms](/glossary/) ## Sources - docs.euno.ai/ai/mcp-setup.md: https://docs.euno.ai/ai/mcp-setup.md (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/glossary/token/ --- # Tool, defined with a sourced example GLOSSARY, read 10 Nov 2026 **The short answer.** A named function an MCP server offers to the model. The model chooses a tool, fills in its arguments, and the server runs it and returns the result. In data servers a tool may read metadata, run a query, or change something. Per AWS Labs' README, the Amazon Redshift MCP server lists 8 tools for discovery, query and cluster review. ## Definition A named function an MCP server offers to the model. The model chooses a tool, fills in its arguments, and the server runs it and returns the result. In data servers a tool may read metadata, run a query, or change something. ## Example from a real server **Example.** Per AWS Labs' README, the Amazon Redshift MCP server lists 8 tools for discovery, query and cluster review. ## Related terms - [tool annotations](/glossary/tool-annotations/): Optional hints a server attaches to a tool, such as readOnlyHint, destructiveHint, idempotentHint and openWorldHint. - [tool definition](/glossary/tool-definition/): The description of one tool that a server sends to the client: its unique name, a title, a text description and an input schema. - [token](/glossary/token/): The unit in which models count text, roughly a word fragment. ## Where to read more - [The directory](/) - [All glossary terms](/glossary/) ## Sources - raw.githubusercontent.com/awslabs/mcp/main/src/redshift-mcp-server/README.md: https://raw.githubusercontent.com/awslabs/mcp/main/src/redshift-mcp-server/README.md (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/glossary/tool/ --- # Tool annotations, defined with a sourced example GLOSSARY, read 10 Nov 2026 **The short answer.** Optional hints a server attaches to a tool, such as readOnlyHint, destructiveHint, idempotentHint and openWorldHint. Clients can use them to decide when to ask a person first. They are hints, not enforced permissions. Per the source of Tableau MCP, the query-datasource tool is annotated readOnlyHint true and destructiveHint false. ## Definition Optional hints a server attaches to a tool, such as readOnlyHint, destructiveHint, idempotentHint and openWorldHint. Clients can use them to decide when to ask a person first. They are hints, not enforced permissions. ## Example from a real server **Example.** Per the source of Tableau MCP, the query-datasource tool is annotated readOnlyHint true and destructiveHint false. ## Related terms - [tool definition](/glossary/tool-definition/): The description of one tool that a server sends to the client: its unique name, a title, a text description and an input schema. - [tool-definition size](/glossary/tool-definition-size/): The amount of text a server's tool definitions take: names, descriptions and input schemas together. - [tool](/glossary/tool/): A named function an MCP server offers to the model. ## Where to read more - [The directory](/) - [All glossary terms](/glossary/) ## Sources - raw.githubusercontent.com/tableau/tableau-mcp/main/src/tools/web/queryDatasource/queryDatasource.ts: https://raw.githubusercontent.com/tableau/tableau-mcp/main/src/tools/web/queryDatasource/queryDatasource.ts (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/glossary/tool-annotations/ --- # Tool definition, defined with a sourced example GLOSSARY, read 10 Nov 2026 **The short answer.** The description of one tool that a server sends to the client: its unique name, a title, a text description and an input schema. The model reads this text to decide when to call the tool. Per the MCP specification, each tool is uniquely identified by a name and includes metadata describing its schema. ## Definition The description of one tool that a server sends to the client: its unique name, a title, a text description and an input schema. The model reads this text to decide when to call the tool. ## Example from a real server **Example.** Per the MCP specification, each tool is uniquely identified by a name and includes metadata describing its schema. ## Related terms - [tool-definition size](/glossary/tool-definition-size/): The amount of text a server's tool definitions take: names, descriptions and input schemas together. - [tools/list](/glossary/tools-list/): The protocol request a client sends to ask a server which tools it offers. - [tool annotations](/glossary/tool-annotations/): Optional hints a server attaches to a tool, such as readOnlyHint, destructiveHint, idempotentHint and openWorldHint. ## Where to read more - [The directory](/) - [All glossary terms](/glossary/) ## Sources - modelcontextprotocol.io/specification/2025-11-25/server/tools: https://modelcontextprotocol.io/specification/2025-11-25/server/tools (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/glossary/tool-definition/ --- # Tool-definition size, defined with a sourced example GLOSSARY, read 10 Nov 2026 **The short answer.** The amount of text a server's tool definitions take: names, descriptions and input schemas together. It is loaded into the model's context on connection, is counted in tokens, and varies widely between servers. Per Alation's README, an enabled-tools list, set by a command-line argument or ALATION_ENABLED_TOOLS, limits which tools the server exposes. ## Definition The amount of text a server's tool definitions take: names, descriptions and input schemas together. It is loaded into the model's context on connection, is counted in tokens, and varies widely between servers. ## Example from a real server **Example.** Per Alation's README, an enabled-tools list, set by a command-line argument or ALATION_ENABLED_TOOLS, limits which tools the server exposes. ## Related terms - [tools/list](/glossary/tools-list/): The protocol request a client sends to ask a server which tools it offers. - [Unity Catalog function as a tool](/glossary/unity-catalog-function-as-a-tool/): A function registered in Databricks Unity Catalog and presented to an agent as an MCP tool. - [tool definition](/glossary/tool-definition/): The description of one tool that a server sends to the client: its unique name, a title, a text description and an input schema. ## Where to read more - [The directory](/) - [All glossary terms](/glossary/) ## Sources - github.com/Alation/alation-ai-agent-sdk/blob/main/README.md: https://github.com/Alation/alation-ai-agent-sdk/blob/main/README.md (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/glossary/tool-definition-size/ --- # Tools/list, defined with a sourced example GLOSSARY, read 10 Nov 2026 **The short answer.** The protocol request a client sends to ask a server which tools it offers. The reply carries every tool definition, and long lists can be split into pages with a cursor. The model sees what comes back. Per Snowflake's docs, MCP clients that connect to a managed MCP server, after authentication, can discover and invoke the tools it specifies. ## Definition The protocol request a client sends to ask a server which tools it offers. The reply carries every tool definition, and long lists can be split into pages with a cursor. The model sees what comes back. ## Example from a real server **Example.** Per Snowflake's docs, MCP clients that connect to a managed MCP server, after authentication, can discover and invoke the tools it specifies. ## Related terms - [Unity Catalog function as a tool](/glossary/unity-catalog-function-as-a-tool/): A function registered in Databricks Unity Catalog and presented to an agent as an MCP tool. - [write tool](/glossary/write-tool/): A tool that creates, changes or deletes something: a table row, a monitor, a report, a tag. - [tool-definition size](/glossary/tool-definition-size/): The amount of text a server's tool definitions take: names, descriptions and input schemas together. ## Where to read more - [The directory](/) - [All glossary terms](/glossary/) ## Sources - docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-mcp: https://docs.snowflake.com/en/user-guide/snowflake-cortex/cortex-agents-mcp (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/glossary/tools-list/ --- # Unity Catalog function as a tool, defined with a sourced example GLOSSARY, read 10 Nov 2026 **The short answer.** A function registered in Databricks Unity Catalog and presented to an agent as an MCP tool. The function's name and description become the tool definition, and Unity Catalog permissions decide who can call it. Per Databricks Labs' README, the Unity Catalog MCP server exposes UC functions, vector search indexes and Genie spaces as tools, inferred at startup. ## Definition A function registered in Databricks Unity Catalog and presented to an agent as an MCP tool. The function's name and description become the tool definition, and Unity Catalog permissions decide who can call it. ## Example from a real server **Example.** Per Databricks Labs' README, the Unity Catalog MCP server exposes UC functions, vector search indexes and Genie spaces as tools, inferred at startup. ## Related terms - [write tool](/glossary/write-tool/): A tool that creates, changes or deletes something: a table row, a monitor, a report, a tag. - [column-level lineage](/glossary/column-level-lineage/): Lineage that follows individual columns, showing which source columns feed which output column. - [tools/list](/glossary/tools-list/): The protocol request a client sends to ask a server which tools it offers. ## Where to read more - [The directory](/) - [All glossary terms](/glossary/) ## Sources - raw.githubusercontent.com/databrickslabs/mcp/master/README.md: https://raw.githubusercontent.com/databrickslabs/mcp/master/README.md (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/glossary/unity-catalog-function-as-a-tool/ --- # Write tool, defined with a sourced example GLOSSARY, read 10 Nov 2026 **The short answer.** A tool that creates, changes or deletes something: a table row, a monitor, a report, a tag. Write tools are where a data agent can do damage, so servers add switches, roles or confirmation around them. Per Google Cloud's docs, the BigQuery tool execute_sql can run INSERT, UPDATE, DELETE and CREATE statements, while execute_sql_readonly is restricted to SELECT. ## Definition A tool that creates, changes or deletes something: a table row, a monitor, a report, a tag. Write tools are where a data agent can do damage, so servers add switches, roles or confirmation around them. ## Example from a real server **Example.** Per Google Cloud's docs, the BigQuery tool execute_sql can run INSERT, UPDATE, DELETE and CREATE statements, while execute_sql_readonly is restricted to SELECT. ## Related terms - [column-level lineage](/glossary/column-level-lineage/): Lineage that follows individual columns, showing which source columns feed which output column. - [context window](/glossary/context-window/): The limit on how much text a model can read at once, including the conversation, tool definitions and every tool result. - [Unity Catalog function as a tool](/glossary/unity-catalog-function-as-a-tool/): A function registered in Databricks Unity Catalog and presented to an agent as an MCP tool. ## Where to read more - [The directory](/) - [All glossary terms](/glossary/) ## Sources - docs.cloud.google.com/bigquery/docs/reference/mcp: https://docs.cloud.google.com/bigquery/docs/reference/mcp (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/glossary/write-tool/ --- # 36 terms, each with an example from a real server GLOSSARY, read 10 Nov 2026 **The short answer.** The glossary defines 36 terms an engineer meets when wiring an agent to data tools over MCP, from tool definition and Streamable HTTP to persona scoping and monitor as code. Each term has a plain definition and an example from one of the 30 servers or the MCP specification, with its source, read 10 Nov 2026. - [column-level lineage](/glossary/column-level-lineage/): Lineage that follows individual columns, showing which source columns feed which output column. - [context window](/glossary/context-window/): The limit on how much text a model can read at once, including the conversation, tool definitions and every tool result. - [Cortex Analyst](/glossary/cortex-analyst/): Snowflake's natural-language analytics service. - [data observability incident](/glossary/data-observability-incident/): A recorded problem found by monitoring, such as late data, a failed test or a volume change. - [Discovery API](/glossary/discovery-api/): The dbt platform API that returns metadata about a project's models, sources, exposures, tests and runs. - [dynamic client registration](/glossary/dynamic-client-registration/): A step where an MCP client registers itself with an authorization server automatically, so nobody has to create a client ID by hand. - [Genie space](/glossary/genie-space/): A Databricks workspace for asking questions of chosen tables in natural language, set up with curated instructions. - [human confirmation](/glossary/human-confirmation/): A step where a person approves a tool call before it runs. - [impact analysis](/glossary/impact-analysis/): The question of what breaks downstream if an asset changes or fails. - [input schema](/glossary/input-schema/): A JSON Schema that states which arguments a tool accepts, their types and which are required or limited to set values. - [local MCP server](/glossary/local-mcp-server/): An MCP server that runs on the engineer's own machine or infrastructure, usually started by the client over stdio. - [MCP client](/glossary/mcp-client/): The part of an AI application that connects to MCP servers, lists their tools and sends the calls the model decides to make. - [MCP server](/glossary/mcp-server/): A program that exposes tools, resources or prompts to an AI application over the Model Context Protocol. - [mcp-remote bridge](/glossary/mcp-remote-bridge/): An open-source Node.js proxy that lets a client which only speaks stdio reach a remote MCP server. - [metadata graph](/glossary/metadata-graph/): A connected store of data assets and their relationships: tables, columns, dashboards, owners and the links between them. - [monitor as code](/glossary/monitor-as-code/): Defining a data monitor as configuration kept in a repository instead of clicking it together in a UI. - [OAuth 2.1 authorization for MCP](/glossary/oauth-2-1-authorization-for-mcp/): The sign-in flow the MCP specification defines for HTTP servers. - [persona scoping](/glossary/persona-scoping/): Limiting what an agent can see or do by the persona of the signed-in person, such as admin, member or guest. - [personal access token](/glossary/personal-access-token/): A long-lived secret a person creates in a tool to let a script or agent act as them. - [read-only tool](/glossary/read-only-tool/): A tool that changes nothing in the system it talks to. - [remote MCP server](/glossary/remote-mcp-server/): An MCP server the vendor hosts and the client reaches by URL. - [resource](/glossary/resource/): Read-only context an MCP server exposes by address, such as a file or a database schema. - [row-level security](/glossary/row-level-security/): A rule that filters which rows a person may read in a table or model, often by role or department. - [semantic layer query](/glossary/semantic-layer-query/): A query written against defined metrics and dimensions instead of raw tables. - [service account](/glossary/service-account/): A non-human identity with its own role, used by unattended agents and jobs. - [stdio transport](/glossary/stdio-transport/): A way to run an MCP server as a local subprocess. - [Streamable HTTP](/glossary/streamable-http/): The MCP transport for servers reached over a network. - [text-to-SQL](/glossary/text-to-sql/): A tool that turns a plain-language question into a SQL statement, usually using what it knows about the schema or project. - [token](/glossary/token/): The unit in which models count text, roughly a word fragment. - [tool](/glossary/tool/): A named function an MCP server offers to the model. - [tool annotations](/glossary/tool-annotations/): Optional hints a server attaches to a tool, such as readOnlyHint, destructiveHint, idempotentHint and openWorldHint. - [tool definition](/glossary/tool-definition/): The description of one tool that a server sends to the client: its unique name, a title, a text description and an input schema. - [tool-definition size](/glossary/tool-definition-size/): The amount of text a server's tool definitions take: names, descriptions and input schemas together. - [tools/list](/glossary/tools-list/): The protocol request a client sends to ask a server which tools it offers. - [Unity Catalog function as a tool](/glossary/unity-catalog-function-as-a-tool/): A function registered in Databricks Unity Catalog and presented to an agent as an MCP tool. - [write tool](/glossary/write-tool/): A tool that creates, changes or deletes something: a table row, a monitor, a report, a tag. Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/glossary/ --- # AWS Labs Redshift MCP server 0.1.3 adds a retrieve_skill_file tool NEWS, 9 NOV 2026, read 10 Nov 2026 **The short answer.** AWS Labs published awslabs.redshift-mcp-server 0.1.3 on PyPI on 9 November 2026. Its README gains retrieve_skill_file, which returns the Redshift guide or a cited reference file in 25,000 character chunks. The wording of MAX_RESULT_ROWS now covers every tool that runs SQL, and those tools take cluster_type with the identifier. Comparing the PyPI project description of 0.1.3 with 0.1.2 shows the documented change. retrieve_skill_file(file, chunk) returns the Redshift guide, or a reference file the guide cites, as markdown. The README says the guide is a copy of the redshift-guide skill from aws/agent-toolkit-for-aws, covering Redshift SQL and how it differs from PostgreSQL, metadata, connectivity and authorization, storage, and query performance. The same description now says a result larger than MAX_RESULT_ROWS, default 1000, is refused by every tool that runs SQL, where 0.1.2 said every tool but list_clusters. It also says every tool that runs SQL takes the cluster_type, provisioned or serverless, with the identifier. The README states the server is read-only by default and that the database user's privileges, not the server, are the security boundary. AWS Labs publishes no separate changelog entry for this version. ## More MCP server news - [Tableau MCP 6.0.4 adds the tableau:entitlements:read OAuth scope](/news/tableau-mcp-6-0-4-entitlements-scope/): 9 Nov 2026 - [Power BI Modeling MCP 1.0.1 connects to workspace semantic models by ID](/news/powerbi-modeling-mcp-1-0-1-workspace-ids/): 6 Nov 2026 - [MCP Toolbox for Databases v1.16.0 adds skills over MCP and Looker LookML dashboard import](/news/mcp-toolbox-1-16-skills-and-looker-dashboards/): 5 Nov 2026 - [Tableau MCP 5.5.0 adds Desktop data app export and knowledge tool switches](/news/tableau-mcp-5-5-desktop-data-apps/): 4 Nov 2026 ## Sources - pypi.org/project/awslabs.redshift-mcp-server/0.1.3: https://pypi.org/project/awslabs.redshift-mcp-server/0.1.3/ (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/news/redshift-mcp-server-0-1-3-skill-file-tool/ --- # Tableau MCP 6.0.4 adds the tableau:entitlements:read OAuth scope NEWS, 9 NOV 2026, read 10 Nov 2026 **The short answer.** Tableau published tableau-mcp 6.0.4 on 9 November 2026. Hosted Tableau MCP now advertises the tableau:entitlements:read OAuth scope, which Pulse insights tools use to confirm Tableau+. Tokens issued earlier must be re-consented under strict scope matching. Desktop invokeCommand and the command tools are removed, a breaking change. The v6.0.4 notes put the scope change first. tableau:entitlements:read is now part of the advertised OAuth scope set, and registration-condition checks run by default. With strict scope matching, which is the default, tokens issued before this release must be re-consented. With DISABLE_STRICT_SCOPE_MATCH=true, existing tokens keep working and the AI Pulse insight tools are left out instead of failing. Connectors that pin a static scope list need to add the new scope. Tool changes: get-datasource-metadata takes optional field selectors, query-datasource accepts internal field names, and skills/list supports pagination. For Desktop, the notes call out two breaking changes: invokeCommand and the command tools are removed, and authoring knowledge is renamed expertise. New Desktop tools are close-workbook and list-saved-datasources. The notes also say staged-upload S3 keys are now scoped per tenant, and the Desktop web profile listener binds to loopback, with a new HTTP_HOST variable for the bind address. ## More MCP server news - [AWS Labs Redshift MCP server 0.1.3 adds a retrieve_skill_file tool](/news/redshift-mcp-server-0-1-3-skill-file-tool/): 9 Nov 2026 - [Power BI Modeling MCP 1.0.1 connects to workspace semantic models by ID](/news/powerbi-modeling-mcp-1-0-1-workspace-ids/): 6 Nov 2026 - [MCP Toolbox for Databases v1.16.0 adds skills over MCP and Looker LookML dashboard import](/news/mcp-toolbox-1-16-skills-and-looker-dashboards/): 5 Nov 2026 - [Tableau MCP 5.5.0 adds Desktop data app export and knowledge tool switches](/news/tableau-mcp-5-5-desktop-data-apps/): 4 Nov 2026 ## Sources - github.com/tableau/tableau-mcp/releases/tag/v6.0.4: https://github.com/tableau/tableau-mcp/releases/tag/v6.0.4 (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/news/tableau-mcp-6-0-4-entitlements-scope/ --- # Power BI Modeling MCP 1.0.1 connects to workspace semantic models by ID NEWS, 6 NOV 2026, read 10 Nov 2026 **The short answer.** Microsoft released powerbi-modeling-mcp 1.0.1 on 6 November 2026, six weeks after 1.0.0. It can connect to semantic models in non-personal workspaces by workspace and semantic model ID rather than names, exposes typed TOM execution, refresh and permission features, and makes dax_query_operations EXECUTE work with query-only connections. The v1.0.1 notes list two features. A client can connect to semantic models in non-personal workspaces by passing the workspace ID and semantic model ID instead of names, and the server exposes typed Tabular Object Model execution, refresh and permission features. The fixes cover reconnects and diagnostics: failed cloud reconnects are retried once with fresh connections, typed connection error diagnostics are preserved without logging server descriptions, and dax_query_operations EXECUTE works with query-only connections. The notes also restore local Power BI Desktop discovery after Windows update KB5127936 on Windows 11 build 26300.9818. The notes remove macOS from the supported OS list for the npm package. The v1.0.0 release on 25 September 2026 changed only the EULA to the GA version, per its own notes. ## More MCP server news - [AWS Labs Redshift MCP server 0.1.3 adds a retrieve_skill_file tool](/news/redshift-mcp-server-0-1-3-skill-file-tool/): 9 Nov 2026 - [Tableau MCP 6.0.4 adds the tableau:entitlements:read OAuth scope](/news/tableau-mcp-6-0-4-entitlements-scope/): 9 Nov 2026 - [MCP Toolbox for Databases v1.16.0 adds skills over MCP and Looker LookML dashboard import](/news/mcp-toolbox-1-16-skills-and-looker-dashboards/): 5 Nov 2026 - [Tableau MCP 5.5.0 adds Desktop data app export and knowledge tool switches](/news/tableau-mcp-5-5-desktop-data-apps/): 4 Nov 2026 ## Sources - github.com/microsoft/powerbi-modeling-mcp/releases/tag/v1.0.1: https://github.com/microsoft/powerbi-modeling-mcp/releases/tag/v1.0.1 (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/news/powerbi-modeling-mcp-1-0-1-workspace-ids/ --- # MCP Toolbox for Databases v1.16.0 adds skills over MCP and Looker LookML dashboard import NEWS, 5 NOV 2026, read 10 Nov 2026 **The short answer.** Google's googleapis/mcp-toolbox v1.16.0 shipped on 5 November 2026. It adds a skills over MCP extension, a Looker import_dashboard_from_lookml tool, support for dynamic UI resources, and a fix making the writeMode protected setting of bigquery-execute-sql fail closed for non-SELECT statements. The Dataplex prebuilt config drops search_aspect_types. The v1.16.0 changelog lists the skills over MCP extension and server support for dynamic UI resources, a ResourceProvider and StructuredToolResult. For Looker it adds import_dashboard_from_lookml and support for text, markdown and rich-text dashboard elements, and it fixes dashboard_filters binding in add_dashboard_element and update_dashboard_element. On BigQuery, the changelog says bigquery-execute-sql now makes writeMode protected fail closed for non-SELECT statements. It also sets UseLegacySql to false when creating a session, recovers default-ADC clients after an on-disk re-auth, and adds debug logging across BigQuery tools. Two operator-facing items: the server now watches tools.yaml for reload when the config flag is omitted, and the Dataplex prebuilt configuration no longer includes the search_aspect_types tool. ## More MCP server news - [AWS Labs Redshift MCP server 0.1.3 adds a retrieve_skill_file tool](/news/redshift-mcp-server-0-1-3-skill-file-tool/): 9 Nov 2026 - [Tableau MCP 6.0.4 adds the tableau:entitlements:read OAuth scope](/news/tableau-mcp-6-0-4-entitlements-scope/): 9 Nov 2026 - [Power BI Modeling MCP 1.0.1 connects to workspace semantic models by ID](/news/powerbi-modeling-mcp-1-0-1-workspace-ids/): 6 Nov 2026 - [Tableau MCP 5.5.0 adds Desktop data app export and knowledge tool switches](/news/tableau-mcp-5-5-desktop-data-apps/): 4 Nov 2026 ## Sources - github.com/googleapis/mcp-toolbox/releases/tag/v1.16.0: https://github.com/googleapis/mcp-toolbox/releases/tag/v1.16.0 (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/news/mcp-toolbox-1-16-skills-and-looker-dashboards/ --- # Tableau MCP 5.5.0 adds Desktop data app export and knowledge tool switches NEWS, 4 NOV 2026, read 10 Nov 2026 **The short answer.** Tableau published tableau-mcp 5.5.0 on 4 November 2026. Changes include scaffolding and exporting workbook data apps from Desktop, a listing of workbook extensions, environment variables to switch the Knowledge tools on or off, a flag giving claude.ai access to MCP Apps tools, and a refusal to apply a workbook change that would delete sheets. The v5.5.0 notes are a list of merged changes. Among them: Desktop can scaffold and export workbook data apps, Desktop lists workbook extensions, and the apply-workbook tool refuses an apply that would delete sheets. Desktop support was synced with External Client API 0.2.28 and 0.2.31. On the hosted and server side, new environment variables turn the Knowledge tools on or off, and a new flag lets claude.ai get MCP Apps tools. The notes also list a change to support scope flexibility, and a fix so get-view-data uses the allData endpoint only when view-all-data is on. An engineer deploying this server can now disable the Knowledge tools by configuration instead of by client-side filtering. Per the notes, product telemetry no longer sends the host name. ## More MCP server news - [AWS Labs Redshift MCP server 0.1.3 adds a retrieve_skill_file tool](/news/redshift-mcp-server-0-1-3-skill-file-tool/): 9 Nov 2026 - [Tableau MCP 6.0.4 adds the tableau:entitlements:read OAuth scope](/news/tableau-mcp-6-0-4-entitlements-scope/): 9 Nov 2026 - [Power BI Modeling MCP 1.0.1 connects to workspace semantic models by ID](/news/powerbi-modeling-mcp-1-0-1-workspace-ids/): 6 Nov 2026 - [MCP Toolbox for Databases v1.16.0 adds skills over MCP and Looker LookML dashboard import](/news/mcp-toolbox-1-16-skills-and-looker-dashboards/): 5 Nov 2026 ## Sources - github.com/tableau/tableau-mcp/releases/tag/v5.5.0: https://github.com/tableau/tableau-mcp/releases/tag/v5.5.0 (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/news/tableau-mcp-5-5-desktop-data-apps/ --- # dbt MCP server 2.10.1 restates job and run access requirements NEWS, 3 NOV 2026, read 10 Nov 2026 **The short answer.** dbt Labs published dbt-mcp 2.10.1 on 3 November 2026. The release rewrites the description of job and run access to use owner targets and shared listing policies instead of unconditional account-wide requirements. Authorization hosts must handle JOBS_LISTING and RUNS_LISTING. It also removes unused full-text product-docs helpers and the index_limits configuration. The v2.10.1 notes list one bug fix and one cleanup. The fix changes how job and run access is described: owner targets and shared listing policies replace the earlier unconditional account-wide requirements. The notes state that authorization hosts must handle JOBS_LISTING and RUNS_LISTING. The cleanup removes unused full-text product-docs helpers and their dedicated index_limits configuration. Per the notes, the product-docs MCP tools still search llms.txt and retrieve individual pages, so the tool surface for an agent is unchanged. For an engineer who runs dbt-mcp behind an authorization layer, the practical item is the two listing names. A host that gates job and run listing needs a rule for each of JOBS_LISTING and RUNS_LISTING. ## More MCP server news - [AWS Labs Redshift MCP server 0.1.3 adds a retrieve_skill_file tool](/news/redshift-mcp-server-0-1-3-skill-file-tool/): 9 Nov 2026 - [Tableau MCP 6.0.4 adds the tableau:entitlements:read OAuth scope](/news/tableau-mcp-6-0-4-entitlements-scope/): 9 Nov 2026 - [Power BI Modeling MCP 1.0.1 connects to workspace semantic models by ID](/news/powerbi-modeling-mcp-1-0-1-workspace-ids/): 6 Nov 2026 - [MCP Toolbox for Databases v1.16.0 adds skills over MCP and Looker LookML dashboard import](/news/mcp-toolbox-1-16-skills-and-looker-dashboards/): 5 Nov 2026 ## Sources - github.com/dbt-labs/dbt-mcp/releases/tag/v2.10.1: https://github.com/dbt-labs/dbt-mcp/releases/tag/v2.10.1 (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/news/dbt-mcp-2-10-1-job-run-access-wording/ --- # dbt MCP server 2.9.0 adds a tool that searches public dbt Mesh models NEWS, 27 OCT 2026, read 10 Nov 2026 **The short answer.** dbt Labs released dbt-mcp 2.9.0 on 27 October 2026. The Discovery toolset gains search_public_mesh_models, which works only in multi-project configurations. The release also adds Python 3.14 support and fixes get_column_lineage, which had always uppercased the column name, and a show tool row limit that misfired on the word limit inside SQL. The dbt Labs release notes for v2.9.0 describe search_public_mesh_models as a new Discovery tool for multi-project setups. It finds public dbt Mesh models across the configured producer projects, confirms the producer project identity needed to write a cross-project ref and a dependencies.yml entry, and answers whether a given warehouse relation is already published as a public model. Three fixes change what an agent sees. get_column_lineage no longer uppercases the column name. The show tool applies its row limit unless the SQL has a LIMIT clause, instead of whenever the word limit appears anywhere, such as in a credit_limit column or a comment. dbt CLI and dbt-codegen calls now run off the event loop, so the server keeps answering other requests while a dbt command runs, and a timed-out call kills its dbt subprocess. For an engineer wiring an agent to a multi-project dbt deployment, the release notes list no new auth or transport change. The new tool applies only where more than one project is configured. ## More MCP server news - [AWS Labs Redshift MCP server 0.1.3 adds a retrieve_skill_file tool](/news/redshift-mcp-server-0-1-3-skill-file-tool/): 9 Nov 2026 - [Tableau MCP 6.0.4 adds the tableau:entitlements:read OAuth scope](/news/tableau-mcp-6-0-4-entitlements-scope/): 9 Nov 2026 - [Power BI Modeling MCP 1.0.1 connects to workspace semantic models by ID](/news/powerbi-modeling-mcp-1-0-1-workspace-ids/): 6 Nov 2026 - [MCP Toolbox for Databases v1.16.0 adds skills over MCP and Looker LookML dashboard import](/news/mcp-toolbox-1-16-skills-and-looker-dashboards/): 5 Nov 2026 ## Sources - github.com/dbt-labs/dbt-mcp/releases/tag/v2.9.0: https://github.com/dbt-labs/dbt-mcp/releases/tag/v2.9.0 (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/news/dbt-mcp-2-9-mesh-model-search/ --- # OpenMetadata 2.0.7 fixes MCP OAuth claim mapping and BASE_PATH routing NEWS, 27 OCT 2026, read 10 Nov 2026 **The short answer.** OpenMetadata 2.0.7 was released on 27 October 2026 with three MCP and authentication fixes. OAuth claim mappings now parse claim names containing a colon, MCP routes and login forms respect a non-root BASE_PATH, and a cleanup filter clears entity and security contexts around MCP, WebSocket and single sign-on requests. The 2.0.7-release notes include a section on the MCP server and authentication. The first fix is that MCP claim mappings failed when a source claim name contained a colon; OAuth claim mappings now parse such names without silently dropping them. The second is that MCP OAuth failed when OpenMetadata ran under a non-root BASE_PATH. The notes say MCP routes and login forms now respect the configured context path, and base URLs and OpenID discovery use the same path. The third concerns isolation: requests outside Jersey could reuse cached identity state from an earlier request. A request cleanup filter now clears entity and security contexts before and after MCP, WebSocket and single sign-on requests. The MCP server ships bundled with the platform release. ## More MCP server news - [AWS Labs Redshift MCP server 0.1.3 adds a retrieve_skill_file tool](/news/redshift-mcp-server-0-1-3-skill-file-tool/): 9 Nov 2026 - [Tableau MCP 6.0.4 adds the tableau:entitlements:read OAuth scope](/news/tableau-mcp-6-0-4-entitlements-scope/): 9 Nov 2026 - [Power BI Modeling MCP 1.0.1 connects to workspace semantic models by ID](/news/powerbi-modeling-mcp-1-0-1-workspace-ids/): 6 Nov 2026 - [MCP Toolbox for Databases v1.16.0 adds skills over MCP and Looker LookML dashboard import](/news/mcp-toolbox-1-16-skills-and-looker-dashboards/): 5 Nov 2026 ## Sources - github.com/open-metadata/OpenMetadata/releases/tag/2.0.7-release: https://github.com/open-metadata/OpenMetadata/releases/tag/2.0.7-release (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/news/openmetadata-2-0-7-mcp-oauth-fixes/ --- # Alation AI Agent SDK and MCP server reach 1.0.0 with a CA trust fix NEWS, 26 OCT 2026, read 10 Nov 2026 **The short answer.** Alation released alation-ai-agent-sdk, alation-ai-agent-langchain and alation-ai-agent-mcp as version 1.0.0 on 26 October 2026, its first stable release. The MCP server now requires certifi 2026.10.19 or later, and its Docker image drops the system pip to remove trust in compromised eMudhra intermediate CAs. Alation's v1.0.0 notes call it the first stable release of the three packages. Requests now send an Ai-Surface header, default sdk, and AgentSDKOptions(surface=...) lets an integration identify itself. Auth parameter objects redact secrets in their repr(), and streaming and non-streaming responses handle malformed or empty payloads more gracefully. The security item concerns the MCP server. It requires certifi 2026.10.19 or later, its Docker image refreshes the OS CA store, and the image no longer ships the base image's system pip. Alation states this removes trust in the compromised eMudhra intermediate CAs. The notes tell anyone running the MCP Docker image, ghcr.io/alation/alation-ai-agent-sdk/alation-mcp-server at v1.0.0, to pull it again to get the CA fix. ## More MCP server news - [AWS Labs Redshift MCP server 0.1.3 adds a retrieve_skill_file tool](/news/redshift-mcp-server-0-1-3-skill-file-tool/): 9 Nov 2026 - [Tableau MCP 6.0.4 adds the tableau:entitlements:read OAuth scope](/news/tableau-mcp-6-0-4-entitlements-scope/): 9 Nov 2026 - [Power BI Modeling MCP 1.0.1 connects to workspace semantic models by ID](/news/powerbi-modeling-mcp-1-0-1-workspace-ids/): 6 Nov 2026 - [MCP Toolbox for Databases v1.16.0 adds skills over MCP and Looker LookML dashboard import](/news/mcp-toolbox-1-16-skills-and-looker-dashboards/): 5 Nov 2026 ## Sources - github.com/Alation/alation-ai-agent-sdk/releases/tag/v1.0.0: https://github.com/Alation/alation-ai-agent-sdk/releases/tag/v1.0.0 (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/news/alation-ai-agent-sdk-1-0-0-stable/ --- # Collibra MCP server chip v0.0.65 returns multi-value attributes in get_asset_details NEWS, 21 OCT 2026, read 10 Nov 2026 **The short answer.** Collibra released chip v0.0.65 on 21 October 2026 with one change to the server: the get_asset_details tool now includes multiValueAttributes in its GraphQL query. Release v0.0.67, published 27 October 2026, contains a changelog diff fix in CI and a Go 1.26.10 bump, so an agent reading assets sees no new tool. The v0.0.65 notes list a single fix: get_asset_details now includes multiValueAttributes in its GraphQL query. An agent that asks Collibra for an asset's details can therefore receive attributes that hold several values. The next published release, v0.0.67 on 27 October 2026, lists two changes: the release job now diffs the changelog against the last published release, and the Go toolchain moves to 1.26.10. Neither changes the tool list. The notes name no change to auth or transport in either release. ## More MCP server news - [AWS Labs Redshift MCP server 0.1.3 adds a retrieve_skill_file tool](/news/redshift-mcp-server-0-1-3-skill-file-tool/): 9 Nov 2026 - [Tableau MCP 6.0.4 adds the tableau:entitlements:read OAuth scope](/news/tableau-mcp-6-0-4-entitlements-scope/): 9 Nov 2026 - [Power BI Modeling MCP 1.0.1 connects to workspace semantic models by ID](/news/powerbi-modeling-mcp-1-0-1-workspace-ids/): 6 Nov 2026 - [MCP Toolbox for Databases v1.16.0 adds skills over MCP and Looker LookML dashboard import](/news/mcp-toolbox-1-16-skills-and-looker-dashboards/): 5 Nov 2026 ## Sources - github.com/collibra/chip/releases/tag/v0.0.65: https://github.com/collibra/chip/releases/tag/v0.0.65 (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/news/collibra-chip-0-0-65-multivalue-attributes/ --- # DataHub MCP server 0.8.0 adds a read-only get_aspect_history tool NEWS, 21 OCT 2026, read 10 Nov 2026 **The short answer.** The acryldata mcp-server-datahub 0.8.0 release on 21 October 2026 adds a read-only get_aspect_history tool. get_entities now returns customProperties, structuredProperties and ownership for all entity types, and list_schema_fields returns curated structured properties per field. All registered write tools are now tagged as mutations. The v0.8.0 notes list get_aspect_history as a new read-only tool. get_entities now returns custom properties, structured properties and ownership for all entity types, exposes Dataset.externalUrl, and requests DataJob.inputOutput so declared job inputs come back. list_schema_fields returns curated structured properties per field, and its pagination continuation was fixed. Write-side changes: all registered write tools are now tagged as mutations, logs distinguish deployment-gated tools, and a documentation change clarifies the mutation gate on save_document. Tool schemas were changed to single-type, and stringified list arguments are coerced. Two limits became configurable: the 120-character field description limit in schema fields and the column query length limit. The notes also say sort_by set to relevance is now a no-op instead of failing. ## More MCP server news - [AWS Labs Redshift MCP server 0.1.3 adds a retrieve_skill_file tool](/news/redshift-mcp-server-0-1-3-skill-file-tool/): 9 Nov 2026 - [Tableau MCP 6.0.4 adds the tableau:entitlements:read OAuth scope](/news/tableau-mcp-6-0-4-entitlements-scope/): 9 Nov 2026 - [Power BI Modeling MCP 1.0.1 connects to workspace semantic models by ID](/news/powerbi-modeling-mcp-1-0-1-workspace-ids/): 6 Nov 2026 - [MCP Toolbox for Databases v1.16.0 adds skills over MCP and Looker LookML dashboard import](/news/mcp-toolbox-1-16-skills-and-looker-dashboards/): 5 Nov 2026 ## Sources - github.com/acryldata/mcp-server-datahub/releases/tag/v0.8.0: https://github.com/acryldata/mcp-server-datahub/releases/tag/v0.8.0 (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/news/datahub-mcp-0-8-0-aspect-history/ --- # Fivetran MCP server 0.5.0 moves credential endpoint control to DISALLOWED_ACTIONS NEWS, 6 OCT 2026, read 10 Nov 2026 **The short answer.** Fivetran released fivetran-mcp 0.5.0 on 6 October 2026. The hardcoded EXCLUDED_ENDPOINTS blocklist is removed, so seven API key endpoints, such as create_system_key and rotate_user_api_key, are now reachable unless DISALLOWED_ACTIONS denies them. Calls other than GET get a 180 second read timeout instead of 30 seconds, and timeouts now say the action may have succeeded. The v0.5.0 notes open with a warning to read the first section before upgrading. The manifest now carries every endpoint in the Fivetran API. Seven endpoints return to reach: create, rotate, update and delete for system keys, plus create_user_api_key, rotate_user_api_key and delete_user_api_keys. Per the notes, DISALLOWED_ACTIONS is now the only thing holding them back, and the server logs a startup warning while any remain callable. Timeouts change too. GET keeps a 30 second read timeout and every other method gets 180 seconds, because setup-test endpoints such as create_connection can run past 30 seconds while Fivetran still applies the write. A timeout now raises UpstreamTimeout with a message saying the action may have succeeded upstream. In streamable-http mode, FIVETRAN_SCOPE, DISALLOWED_ACTIONS and FIVETRAN_ALLOW_WRITES now work as they do in stdio, including the read-only default when nothing is set. ## More MCP server news - [AWS Labs Redshift MCP server 0.1.3 adds a retrieve_skill_file tool](/news/redshift-mcp-server-0-1-3-skill-file-tool/): 9 Nov 2026 - [Tableau MCP 6.0.4 adds the tableau:entitlements:read OAuth scope](/news/tableau-mcp-6-0-4-entitlements-scope/): 9 Nov 2026 - [Power BI Modeling MCP 1.0.1 connects to workspace semantic models by ID](/news/powerbi-modeling-mcp-1-0-1-workspace-ids/): 6 Nov 2026 - [MCP Toolbox for Databases v1.16.0 adds skills over MCP and Looker LookML dashboard import](/news/mcp-toolbox-1-16-skills-and-looker-dashboards/): 5 Nov 2026 ## Sources - github.com/fivetran/fivetran-mcp/releases/tag/v0.5.0: https://github.com/fivetran/fivetran-mcp/releases/tag/v0.5.0 (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/news/fivetran-mcp-0-5-0-disallowed-actions/ --- # 12 MCP server releases, each from its primary source NEWS, read 10 Nov 2026 **The short answer.** Twelve dated releases of MCP servers for data tools, from 6 Oct 2026 to 9 Nov 2026, each from its primary source: GitHub release pages, package registries and vendor release notes. Each item says what shipped (new tools, auth scopes, transport or access changes) and what it means for an engineer wiring an agent to that tool. - [AWS Labs Redshift MCP server 0.1.3 adds a retrieve_skill_file tool](/news/redshift-mcp-server-0-1-3-skill-file-tool/): 9 Nov 2026 - [Tableau MCP 6.0.4 adds the tableau:entitlements:read OAuth scope](/news/tableau-mcp-6-0-4-entitlements-scope/): 9 Nov 2026 - [Power BI Modeling MCP 1.0.1 connects to workspace semantic models by ID](/news/powerbi-modeling-mcp-1-0-1-workspace-ids/): 6 Nov 2026 - [MCP Toolbox for Databases v1.16.0 adds skills over MCP and Looker LookML dashboard import](/news/mcp-toolbox-1-16-skills-and-looker-dashboards/): 5 Nov 2026 - [Tableau MCP 5.5.0 adds Desktop data app export and knowledge tool switches](/news/tableau-mcp-5-5-desktop-data-apps/): 4 Nov 2026 - [dbt MCP server 2.10.1 restates job and run access requirements](/news/dbt-mcp-2-10-1-job-run-access-wording/): 3 Nov 2026 - [dbt MCP server 2.9.0 adds a tool that searches public dbt Mesh models](/news/dbt-mcp-2-9-mesh-model-search/): 27 Oct 2026 - [OpenMetadata 2.0.7 fixes MCP OAuth claim mapping and BASE_PATH routing](/news/openmetadata-2-0-7-mcp-oauth-fixes/): 27 Oct 2026 - [Alation AI Agent SDK and MCP server reach 1.0.0 with a CA trust fix](/news/alation-ai-agent-sdk-1-0-0-stable/): 26 Oct 2026 - [Collibra MCP server chip v0.0.65 returns multi-value attributes in get_asset_details](/news/collibra-chip-0-0-65-multivalue-attributes/): 21 Oct 2026 - [DataHub MCP server 0.8.0 adds a read-only get_aspect_history tool](/news/datahub-mcp-0-8-0-aspect-history/): 21 Oct 2026 - [Fivetran MCP server 0.5.0 moves credential endpoint control to DISALLOWED_ACTIONS](/news/fivetran-mcp-0-5-0-disallowed-actions/): 6 Oct 2026 Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/news/ --- # 30 questions about MCP servers for data, answered with names and dates FAQ, read 10 Nov 2026 **The short answer.** Thirty questions engineers ask about MCP servers for data tools, answered with names, versions and dates read on 10 Nov 2026: which of Snowflake, Databricks, BigQuery, dbt, Tableau and the catalogs publish official servers, remote versus local, which servers can write, how auth works, and how this directory reads each entry. **What is an MCP server for a data tool?** It is a program that exposes a data tool's functions to an AI agent over the Model Context Protocol. The agent lists the tools, then calls them to search metadata, read lineage or run queries. This directory lists 30 such servers for warehouses, dbt, ingestion connectors, BI tools, catalogs and observability tools, each read on 10 Nov 2026. **What is the difference between a remote and a local MCP server?** A remote server is hosted by the vendor and reached by URL, usually with OAuth. A local server runs on your machine, usually over stdio, with credentials in environment variables. Of the 30 servers here, 15 are remote only, 12 offer both, and three run over stdio only: Redshift (AWS Labs), Sifflet and a community Snowflake server. **Which transports do data MCP servers use?** Two: stdio for a process the client starts locally, and Streamable HTTP for a server reached over a network. Fivetran's server runs over stdio by default and as a hosted HTTP server with --transport streamable-http. BigQuery's remote server at Google Cloud uses Streamable HTTP. A client that only speaks stdio can reach a remote server through the mcp-remote bridge. **How does authentication work for data MCP servers?** It varies by server and the directory records each one. Hosted servers such as Monte Carlo use OAuth 2.1 with Dynamic Client Registration, so calls run as the signed-in user. Others take an API token: Select Star's server uses a Bearer header, and Fivetran's uses one shared API key and secret. Redshift uses AWS credentials. **Which data MCP servers can write, not only read?** Read-only entries are few. Select Star's server is read-only with 7 tools. Cube, Hex, OpenMetadata, Sifflet, Sigma and ThoughtSpot are read-write. Sifflet's two state-changing tools open and close incidents. Redshift is read-only by default until ACCESS_MODE is changed, and Airbyte's read-only mode hides run and cancel sync tools. **What does official mean in this directory?** Official means the data tool's own vendor publishes the server, such as dbt Labs for dbt-mcp or Google Cloud for the BigQuery remote server. The directory also marks Labs entries, like Snowflake Labs and Databricks Labs, and community entries. One of the 30 is community: isaacwasserman's Snowflake server, version 0.4.0, last released 10 Apr 2025. **How are entries in this directory read and dated?** Each entry comes from the server's own repository or documentation, with the version or commit and the date it was read. This edition was read on 10 Nov 2026. A vendor's claim is written as the vendor's, and where nothing was found the entry says none found on 10 Nov 2026. **What does the Context Weight Index measure?** It measures the size of each server's published tool definitions in tokens, counted one way for every server. Tool definitions are loaded into the model's context window when a client connects, before any call. The index states a size and nothing else, and it is read from each server's published tool text. **Why does the directory list servers and not order them by merit?** Servers differ in what they cover, how they authenticate and what they can change, and those differences do not reduce to one order. The directory lists the 30 servers alphabetically and puts the read facts side by side. The reader decides which facts matter for the situation at hand. **Does Snowflake publish an official MCP server?** Yes. Snowflake documents a Snowflake-managed MCP server, remote, with no package version; the docs page follows MCP revision 25 Nov 2025. It offers Cortex Agent, Cortex Search, Cortex Analyst and SQL execution tools. Snowflake Labs' older server, version 1.4.2 from 15 May 2026, is marked deprecated in its README. A community stdio server also exists. **Does Databricks publish an official MCP server?** Yes. Databricks documents Databricks-provided MCPs in the system.ai schema, remote, with no version published; the docs page was last updated Nov 9, 2026. It lists four: Genie One, dbsql, web_search and sandbox. Databricks Labs' Unity Catalog server has no tagged release, and its README marks it deprecated in favor of the managed servers. **Does BigQuery have an official MCP server?** Yes. Google Cloud documents a BigQuery remote MCP server at bigquery.googleapis.com/mcp over Streamable HTTP, with nine tools and no version number; the reference page shows Last updated 3 Nov 2026. Google's MCP Toolbox for Databases (version v1.16.0, released 5 Nov 2026) also has BigQuery tools, over stdio or HTTP. **Does Amazon Redshift have an official MCP server?** Yes. AWS Labs publishes the Amazon Redshift MCP Server in its awslabs/mcp repository, version 0.1.3, uploaded 9 Nov 2026. It runs over stdio, lists 8 tools and uses AWS credentials. It is read-only by default; read-write mode needs ACCESS_MODE, and each write then asks for confirmation unless that prompt is switched off. **Does dbt publish an official MCP server?** Yes. dbt Labs publishes dbt-mcp, version 2.10.1, released 3 Nov 2026, with 52 tools. It can be self-hosted or used remotely. Its README lists Semantic Layer, Discovery, lineage, text_to_sql and dbt CLI tools, and warns that CLI tools can modify models and warehouse objects, so run it only with a client you trust. **Does Fivetran publish an official MCP server?** Yes. Fivetran publishes fivetran-mcp, version v0.5.1, released 6 Oct 2026. It runs over stdio by default and can run as a hosted HTTP server. Auth is one shared Fivetran API key and secret, not per-user OAuth. Its README says write and delete confirmation is advisory only, and the server does not enforce it. **Does Airbyte publish an official MCP server?** Yes. Airbyte documents a hosted Airbyte MCP at mcp.airbyte.com/mcp, remote, with no versioned release, listing 50 tools. It uses OAuth as the signed-in Airbyte user. PyAirbyte 0.71.3, uploaded 26 Oct 2026, also ships an experimental local airbyte.mcp server. The X-MCP-Readonly-Mode header hides run and cancel sync tools. **Does Tableau publish an official MCP server?** Yes. Tableau (Salesforce) publishes tableau-mcp, version 6.0.4, released 9 Nov 2026, with 40 tools. It speaks stdio by default and HTTP as an alternative. Hosted use signs in with OAuth 2.1 as each user's Tableau Cloud identity. Self-hosted use takes a personal access token by default, or direct trust, OAuth, passthrough or unified access tokens. **Does Looker have an official MCP server?** Looker tools ship inside Google's MCP Toolbox for Databases, version v1.16.0, released 5 Nov 2026. The Toolbox runs as a local stdio server through npx @toolbox-sdk/server, or over HTTP. The directory counts 55 Looker tools. Auth is a Looker API client_id and client_secret, or use_client_oauth to pass the caller's token; the default is false. **Does Power BI have an official MCP server?** Yes. Microsoft publishes the Power BI Authoring MCP Server, version 1.0.1, released 6 Nov 2026, with 21 tools. The local server uses stdio and Entra sign-in or a service principal; a hosted preview endpoint uses Streamable HTTP with Entra ID. The --readonly option blocks writes to the semantic model, and read-write is the default. **Does ThoughtSpot publish an official MCP server?** Yes. ThoughtSpot publishes the Spotter MCP Server in thoughtspot/mcp-server, version 0.5.0, with its last release dated 6 Jun 2025. It is remote and lists 20 tools. Auth is OAuth to a ThoughtSpot instance, with dynamic client registration or a manually registered client ID and secret. Bearer-token and /sse paths also exist. **Does Sigma publish an official MCP server?** Yes. Sigma documents a Sigma-hosted MCP server, remote over HTTP, with no versioned release; the URL is copied from Profile > Integrations. It searches, describes and queries data and builds workbooks. OAuth inherits the user's Sigma permissions, and an API-key path with scope mcp:access is labeled Beta. Its page lists no individual tool names. **Does Omni publish an official MCP server?** Yes. Omni documents a hosted Omni MCP Server, remote, with no versioned release, listing 26 tools. It is enabled by an organization setting under AI Hub > MCP, on by default. Auth is OAuth 2.1 with PKCE, or an API key. An Organization API key can act as another user through the X-MCP-User-ID header. **Does Cube publish an official MCP server?** Yes. Cube documents a hosted Cube MCP server, remote, with no versioned release, listing 30 tools and read-write access. Auth is OAuth Authorization Code with PKCE, or a Cube API key exchanged for a token. Cube says every tool runs as the authenticated user, with row-level security applied, and that nine tools are annotated destructive. **Does Hex publish an official MCP server?** Yes. Hex documents a hosted Hex MCP server at app.hex.tech/mcp, remote, with no versioned release, listing 18 tools and read-write access. Auth is OAuth to the user's workspace. Knowledge tools need the Explorer role or above, and project editing tools need Editor or above. Single-tenant, EU and HIPAA workspaces use their own host. **Which data catalogs publish official MCP servers?** Alation (alation-ai-agent-mcp 1.0.0, released 26 Oct 2026, stdio and remote), Atlan (hosted, unversioned, 41 tools, remote), Collibra (chip v0.0.67, released 27 Oct 2026, both) and DataHub (0.8.0, released 21 Oct 2026, both). OpenMetadata ships 2.0.7, released 27 Oct 2026, remote. Alation's remote server does not support dynamic client registration. **What do Secoda and Select Star publish for MCP?** Secoda publishes secoda-mcp on GitHub: no releases, main branch last commit 26 Jan 2026, stdio and hosted options, authenticated with one Secoda API token. Select Star documents a hosted, remote, read-only MCP server with 7 tools and no version published, authenticated with an API token as a Bearer header; Snowflake announced on 24 Nov 2025 that it agreed to acquire Select Star. **Which observability tools publish official MCP servers?** Monte Carlo runs a hosted server at mcp.getmontecarlo.com/mcp, remote, with no version published and 15 tools. Sifflet publishes sifflet-mcp 0.1.7, released 3 Aug 2026, over stdio. SYNQ (docs titled Coalesce Quality) runs a hosted remote server with no version published and 134 listed tools, using OAuth 2 with PKCE and permission tiers. **What does Euno's MCP server publish?** Euno documents a hosted server at api.app.euno.ai/mcp, remote, with no version published, listing 56 tools. Auth is OAuth sign-in or an API key, and the active persona scopes data and actions. Euno's server gives an agent governed context about the data; it does not run the query, which is the job of the warehouse's own server. **Is there a community MCP server for Snowflake?** Yes. isaacwasserman's mcp-snowflake-server, version 0.4.0, last released 10 Apr 2025, is the one community entry among the 30. It runs over stdio, lists 8 tools including read_query, write_query and create_table, and uses Snowflake user credentials. Snowflake's own managed server and the deprecated Snowflake Labs server are listed separately. **How do I report a correction to an entry?** Use the corrections page on this site. Name the server, the field and the source that shows the change. Each entry carries a version or commit and a read date, so say which one you checked and the date you read the source. That lets the entry be compared with what you saw. Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/faq/ --- # Three datasets, CSV and JSON, free to reuse with a citation DATASETS, read 10 Nov 2026 **The short answer.** The Data MCP Directory publishes three datasets under CC BY 4.0, read 10 Nov 2026: the directory (30 servers, fourteen fields each), the evidence (522 facts, each with its source URL and read date) and the Context Weight Index (24 servers measured in tokens). | Dataset | Rows | CSV | JSON | | --- | --- | --- | --- | | The Data MCP Directory | 30 | data-mcp-directory.csv | data-mcp-directory.json | | The Data MCP Directory evidence | 522 | data-mcp-evidence.csv | data-mcp-evidence.json | | The Context Weight Index | 30 | context-weight-index.csv | context-weight-index.json | ## Cite this Data MCP Directory, datamcpservers.com, edition of 10 Nov 2026, CC BY 4.0. Each JSON file carries the edition, the read date, the license and the cite line. Rows are filed by data tool, then by server name. ## Sources - Method: https://datamcpservers.com/method/ (read 10 Nov 2026) - Context Weight Index methodology: https://datamcpservers.com/context-weight/methodology/ (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/datasets/ --- # The Data MCP Directory reads each MCP server for a data tool from its own source ABOUT THE DATA MCP DIRECTORY, read 10 Nov 2026 **The short answer.** The Data MCP Directory is a publication about MCP servers for data tools. It lists 30 servers for 26 data tools, each read from its own repository or documentation with the version and the read date, measures their tool definitions in the Context Weight Index, and orders nothing by merit. The directory covers one subject: the MCP servers that connect agents to warehouses, dbt, ingestion connectors, BI tools, catalogs and observability tools. Each page answers one question in its headline and carries its sources at the foot. It is written for the engineer wiring an agent to a warehouse who writes down what each server really does: who publishes it, how it signs in, what it can change, and how much of the context window its tool definitions take. The method is on the Method page. Corrections go to the corrections page and are logged with the date and the fact that changed. Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/about/ --- # Every entry is read from the server's own source, labelled by one rule and dated EDITORIAL POLICY, read 10 Nov 2026 **The short answer.** Every entry in the Data MCP Directory is read from the server's own repository, registry page or documentation, labelled by the rules on the Method page and dated. Publisher claims are written as the publisher's. Facts looked for and not found say none found on 10 Nov 2026. Nothing is ordered by merit. Entries are read from primary sources only: the server's repository, its package registry page and the vendor's documentation. Secondary sources appear only as labelled third-party facts. Every server gets the same fourteen fields in the same order. Euno's server, like every other, is read and written the same way. Lists are filed alphabetically by data tool, then by server name. The site orders nothing by merit and names no choice. Entries are re-read at least every 60 days; the read date on each entry says when it was last read. ## Sources - Method: https://datamcpservers.com/method/ (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/editorial-policy/ --- # Report a factual error with its source, and the entry is re-read and the change logged CORRECTIONS, read 10 Nov 2026 **The short answer.** To correct an entry in the Data MCP Directory, send the page address, the fact that is wrong and a source that shows the right fact to editor@datamcpservers.com. The entry is re-read from the server's own source, changed if the source confirms it, and the change is logged here with its date. Corrections are made from sources, not from requests. A publisher's correction is welcome, and is checked against its own repository or documentation like any other. No corrections have been logged since the edition of 10 Nov 2026. ## Sources - Method: https://datamcpservers.com/method/ (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/corrections/ --- # Write to the editors by email, at the address below CONTACT, read 10 Nov 2026 **The short answer.** The Data MCP Directory reads mail at editor@datamcpservers.com. Send corrections with the page address and a source, servers that should be listed with a link to their own repository or documentation, and questions about the Context Weight Index method. Every listed server is read from its own source before it appears. The address is editor@datamcpservers.com. A new server is listed once its own repository or documentation can be read and it serves a data tool. Corrections are handled as the corrections page describes. Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/contact/ --- # No cookies, no analytics scripts, and server logs kept for traffic counts only PRIVACY, read 10 Nov 2026 **The short answer.** The Data MCP Directory sets no cookies and loads no analytics or advertising scripts. The hosting provider's standard access logs record each request (address, time, page, user agent and referrer) and are used only to count page views and crawler visits. Mail sent to the editors is used only to answer it. The only script on the site sorts the Context Weight Index table in the browser; it sends nothing anywhere. Access logs are kept in the site's own storage and are not shared. ## Sources - Method: https://datamcpservers.com/method/ (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/privacy/ --- # The pages and datasets may be reused under CC BY 4.0, with a citation TERMS OF USE, read 10 Nov 2026 **The short answer.** The Data MCP Directory's pages and datasets may be reused under the Creative Commons Attribution 4.0 license, with a citation to the edition used. Entries record what each server's own source said on the read date; check the source before relying on a fact, because servers change between releases. Product names belong to their owners and are used to identify the servers listed. Links to servers' repositories and documentation go to their publishers' own pages. Each entry carries the date it was read. A server can change its tools, its auth or its transport in any release, so a fact on this site describes the version named on the entry and no later one. The Context Weight Index counts tokens with one public encoding. Other tokenizers give other absolute numbers, and the methodology page says how the count is made so that anyone can repeat it. ## Sources - Method: https://datamcpservers.com/method/ (read 10 Nov 2026) Last reviewed 10 Nov 2026. Canonical: https://datamcpservers.com/terms/