Skip to content
Databricks Managed MCP

Databricks Managed MCP

by Databricks
Proprietary (Databricks platform feature)
Repository Docs
Databases & Data
databricksunity-cataloggenievector-searchsqldata-governancelakehouse

Summary

Databricks-hosted MCP endpoints that expose Unity Catalog functions, Genie spaces, Vector Search indexes and Databricks SQL to agents, with UC permissions enforced.

Features

  • Databricks hosts the servers — no deployment or credential plumbing
  • Unity Catalog permissions enforced on every call
  • Separate endpoints for Genie spaces, AI Search indexes, SQL and UC functions
  • Scope an agent to exactly one catalog, schema, index or Genie space
  • On-behalf-of-user OAuth with scopes mirroring the server names
  • Central monitoring through Unity AI Gateway
  • Billed as compute consumed, with no separate per-server charge

Installation

Set up this MCP server in your favorite AI agent — copy a ready-made configuration below.

Any MCP-compatible agent

Most agents (Claude, Cursor, Windsurf, VS Code, and more) read a standard mcpServers configuration.

  1. Open your agent's MCP configuration file.
  2. Merge the snippet below into it.
  3. Restart the agent — the "Databricks Managed MCP" tools will be available.
{
  "mcpServers": {
    "databricks-managed-mcp": {
      "command": "https://<workspace-hostname>/api/2.0/mcp/genie"
    }
  }
}

Description

Databricks Managed MCP is a set of MCP endpoints that Databricks hosts inside your own workspace. There is nothing to deploy: the servers run on the Databricks side, authentication is managed for you, and — the part that matters for a data platform — Unity Catalog permissions are always enforced, so an agent can only reach what the calling identity was already entitled to reach.

The endpoints

Each server type is a distinct URL under your workspace host:

| Server | URL |
| --- | --- |
| Genie (all spaces) | https://<workspace-hostname>/api/2.0/mcp/genie |
| Genie agent (one space) | https://<workspace-hostname>/api/2.0/mcp/genie/{genie_space_id} |
| Vector / AI Search index | https://<workspace-hostname>/api/2.0/mcp/ai-search/{catalog}/{schema}/{index_name} |
| Databricks SQL | https://<workspace-hostname>/api/2.0/mcp/sql |
| Unity Catalog function | https://<workspace-hostname>/api/2.0/mcp/functions/{catalog}/{schema}/{function_name} |

That shape is deliberate: instead of one server with a sprawling tool list, you attach exactly the catalog, schema, index or Genie space an agent should see. A support agent gets one Genie space; an analytics agent gets SQL plus two UC functions.

Authentication and governance

On-behalf-of-user authentication uses OAuth scopes that mirror the server names — genie, ai-search, sql and unity-catalog. Traffic can be monitored centrally through Unity AI Gateway, which is how you get an audit trail of what agents asked your warehouse.

Cost and status

There is no per-server fee; you pay for the compute each call triggers. Unity Catalog functions bill against serverless general compute, Genie agents against serverless SQL compute, the SQL server against SQL compute, and AI Search against vector-search pricing. The managed servers are in preview. Note that the older self-hosted Unity Catalog server in databrickslabs/mcp is deprecated in favour of these managed endpoints.

Who it is for

Teams already standardised on Databricks who want agents querying governed data instead of a copied extract — and who need the permission model to hold without writing a bespoke authorisation layer in front of the agent.

Related MCP Servers

MCP: deja-vu

by Vladislav Shulcz

New

Indexes the coding-agent sessions already sitting on your disk — months of history from before you installed it — and serves them back over MCP. 3.5 GB searched in ~1.5 ms, no LLM, no embeddings.

Developer ToolsDatabases & Data
1 views

MCP: Weaviate MCP Server

by Weaviate

New

Weaviate's MCP server, built into the database itself — hybrid search, schema inspection, tenant listing and batch upserts at /v1/mcp, scoped by RBAC.

Databases & Data

MCP: MCP Toolbox for Databases

by Google

New

Google's open-source MCP server for enterprise databases — one binary that fronts AlloyDB, BigQuery, Spanner, Postgres, MySQL, Oracle, MongoDB, Snowflake and more.

Databases & DataCloud & Infrastructure
3 views

MCP: DBHub

by Bytebase

Featured

A universal database MCP server — one integration for PostgreSQL, MySQL, SQL Server, SQLite, and MariaDB.

Databases & Data
Browse all MCP servers →