CONTEXT WEIGHT INDEX, read 10 Nov 2026
The published tool definitions of 24 MCP servers measure from 63 to 4,550 tokens
Download the indexHow it is measured
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) | 50 | 50 | 870 | 17 | docs page |
| Alation | Alation AI Agent SDK MCP server | 4 | 3 | 69 | 17 | docs page (parameters published, not counted) |
| Amazon Redshift | Amazon Redshift MCP Server (AWS Labs) | 8 | 8 | 238 | 30 | README (parameters published, not counted) |
| Atlan | Atlan MCP Server | 42 | 42 | 1,770 | 42 | docs page (parameters published, not counted) |
| BigQuery | BigQuery remote MCP server | 9 | 9 | 1,487 | 165 | docs page |
| BigQuery | MCP Toolbox for Databases (BigQuery tools) | 9 | 9 | 156 | 17 | docs page |
| Collibra | Collibra MCP Server (chip) | 41 | 41 | 2,914 | 71 | README |
| Cube | Cube MCP server | 30 | 30 | 1,093 | 36 | docs page |
| Databricks | Databricks managed MCP servers | 4 | 4 | 94 | 24 | docs page |
| DataHub | DataHub MCP Server | 22 | 22 | 567 | 26 | README |
| dbt | dbt MCP Server | 63 | 63 | 1,061 | 17 | README |
| Euno | Euno MCP | 56 | 42 | 760 | 14 | docs page |
| Fivetran | Fivetran MCP Server | 53 | 2 | 299 | 6 | README |
| Hex | Hex MCP server | 18 | 18 | 252 | 14 | docs page |
| Looker | Looker tools in MCP Toolbox for Databases | 55 | 55 | 1,475 | 27 | docs index |
| Omni | Omni MCP Server | 27 | 27 | 466 | 17 | docs page |
| OpenMetadata | OpenMetadata MCP Server | 15 | 15 | 1,249 | 83 | docs page (parameters published, not counted) |
| Power BI | Power BI Modeling MCP Server | 21 | 21 | 387 | 18 | README |
| Secoda | Secoda MCP | 6 | 6 | 63 | 11 | source code (parameters published, not counted) |
| Select Star | Select Star MCP Server | 7 | 7 | 98 | 14 | docs page |
| Sifflet | Sifflet MCP Server | 10 | 10 | 662 | 66 | source code (parameters published, not counted) |
| Snowflake | mcp-snowflake-server (community) | 8 | 8 | 103 | 13 | README (parameters published, not counted) |
| SYNQ | SYNQ MCP (Coalesce Quality) | 134 | 134 | 4,550 | 34 | docs page |
| ThoughtSpot | ThoughtSpot MCP Server | 20 | 20 | ≥ 2,149 | 107 | source code (parameters published, not counted) |
Servers that could not be measured
- Databricks Labs MCP servers: 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: no published tool-definition text was read on 10 Nov 2026
- Sigma MCP server: 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: 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: 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: 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: 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, read 10 Nov 2026
- Alation AI Agent SDK MCP server: 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, read 10 Nov 2026
- Atlan MCP Server: 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, 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, read 10 Nov 2026
- Collibra MCP Server (chip): github.com/collibra/chip/blob/main/README.md, read 10 Nov 2026
- Cube MCP server: 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, read 10 Nov 2026
- DataHub MCP Server: 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, read 10 Nov 2026
- Euno MCP: docs.euno.ai/ai/mcp-setup.md, read 10 Nov 2026
- Fivetran MCP Server: raw.githubusercontent.com/fivetran/fivetran-mcp/main/README.md, read 10 Nov 2026
- Hex MCP server: learn.hex.tech/docs/api-integrations/mcp-server, read 10 Nov 2026
- Looker tools in MCP Toolbox for Databases: mcp-toolbox.dev/llms.txt, read 10 Nov 2026
- Omni MCP Server: 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, read 10 Nov 2026
- Power BI Modeling MCP Server: 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, read 10 Nov 2026
- Select Star MCP Server: 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, read 10 Nov 2026
- mcp-snowflake-server (community): raw.githubusercontent.com/isaacwasserman/mcp-snowflake-server/main/README.md, read 10 Nov 2026
- SYNQ MCP (Coalesce Quality): 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, read 10 Nov 2026
Last reviewed 10 Nov 2026.