SIDE BY SIDE, read 10 Nov 2026
Databricks Labs MCP servers and Databricks managed MCP servers, field by field: 3 and 4 tools named
Databricks Labs MCP serversDatabricks managed MCP servers
| Field | Databricks Labs MCP servers | Databricks managed MCP servers |
|---|---|---|
| Data tool | Databricks | Databricks |
| Publisher | Databricks Labs | Databricks |
| Kind | Official, Labs org | Official |
| Repository or docs | github.com/databrickslabs/mcp | 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 | 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 |
In one sentence each
- 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.
- 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.
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 <workspace-hostname>/ai-gateway/mcp-services/<mcp-name>. |
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
Sources
- raw.githubusercontent.com/databrickslabs/mcp/master/README.md, read 10 Nov 2026
- docs.databricks.com/aws/en/generative-ai/mcp/managed-mcp, read 10 Nov 2026
Last reviewed 10 Nov 2026.