Genie MCP Server

What is Genie MCP Server?

Genie MCP Server is an async server that integrates with Databricks Genie via the MCP StreamableHTTP Protocol. It enables users to send conversational queries to Genie and receive answers or data, using Databricks authentication with automatic OAuth token management.

How to use Genie MCP Server?

Deploy the server as a Databricks app by cloning the repository, configuring the Genie space ID in app.yaml, and deploying via the Databricks Apps interface. Once running, send MCP requests to the app’s endpoint.

Key features of Genie MCP Server

• Async API server built with Starlette and Uvicorn
• Integration with Databricks Genie for conversational queries
• Automatic OAuth token management for Databricks
• Deployable as a Databricks app using app.yaml configuration
• Exposes Genie functionality via MCP StreamableHTTP Protocol

Use cases of Genie MCP Server

• Query Databricks Genie for answers to natural language questions
• Retrieve structured data from Genie spaces within Databricks
• Integrate Genie-powered conversational access into MCP-compatible clients
• Deploy a serverless Genie interface inside a Databricks workspace

FAQ from Genie MCP Server

What is the runtime environment required?

The server runs on Python and uses Starlette and Uvicorn. It is designed to be deployed as a Databricks app.

How does authentication work?

Authentication is handled via automatic OAuth token management for Databricks, managed by the token_minter.py module.

Where is the Genie space ID configured?

The Genie space ID is set as the SPACE_ID environment variable in the app.yaml deployment configuration file.

Can I deploy this outside of Databricks?

The README only describes deployment as a Databricks app. No alternative deployment methods are documented.

What transport protocol does the server use?

The server exposes its tool via the MCP StreamableHTTP Protocol.