Databricks Genie API MCP Server

This project implements a Model Context Protocol (MCP) server that exposes Databricks Genie API capabilities as tools. It allows you to integrate Databricks' no-code AI/BI assistant features with other applications through a standardized interface, enabling powerful natural language querying of your Databricks data.

For a detailed explanation and example use cases, please see the accompanying blog post.

Features

• Expose Databricks Genie API functions as MCP tools.
• Enable natural language querying of Databricks data.
• Start and manage Genie conversations.
• Create and retrieve messages.
• Execute and fetch SQL query results generated by Genie.
• Secure authentication with Databricks.

Prerequisites

• Python 3.10+
• Databricks workspace with Genie access and System Tables enabled (if using them).
• Databricks Assistant enabled.
• CAN USE permission on a Pro or Serverless SQL warehouse.
• Access to Unity Catalog data relevant to your Genie Space.
• An MCP-compatible client application, such as Claude Desktop.

Setup Instructions

1. Clone the Repository (if you haven't already)

   # Add clone command if needed
   

2. Navigate to the Server Directory

   cd genie_api 
   

3. Install Dependencies

   pip install -r requirements.txt
   

4. Configure Authentication

   Set the required environment variables for the Databricks SDK to connect to your workspace. Create a .env file in this directory (genie_api/) or set them globally:

   # --- .env file content ---
   
   # For PAT authentication (recommended for development)
   # DATABRICKS_HOST=https://your-workspace.cloud.databricks.com
   # DATABRICKS_TOKEN=your-personal-access-token
   
   # Or for OAuth with service principal (recommended for production)
   DATABRICKS_HOST=https://your-workspace.cloud.databricks.com
   DATABRICKS_CLIENT_ID=your-client-id
   DATABRICKS_CLIENT_SECRET=your-client-secret
   
   # --- end .env file content ---
   

   Ensure the .env file is added to your `.gitignore!

5. Run the Server Locally

   python server.py
   

   The server will start and listen for connections via standard input/output (stdio).

Using with Claude Desktop

This MCP server is designed to be used with MCP clients like Claude Desktop. Follow these steps to connect:

1. Install Claude Desktop: Download and install from the official website.

2. Configure Claude Desktop:

   • Open Claude Desktop settings (Menu Bar -> Claude -> Settings...).
   • Go to Developer -> Edit Config.
   • Add the following entry to the mcpServers object in the claude_desktop_config.json file, adjusting paths as needed:

   {
     "mcpServers": {
       "databricks-genie": {
         "command": "python", // Or python3, or the full path to your python executable
         "args": [
           "/full/absolute/path/to/your/project/genie_api/server.py" 
         ],
         "workingDirectory": "/full/absolute/path/to/your/project/genie_api/" 
       }
       // ... potentially other servers ...
     }
   }
   

   • Important: Use the full absolute path to server.py and the genie_api directory.
   • Ensure the Python command is accessible by Claude Desktop.
   • The workingDirectory ensures the server can find auth.py and your .env file.

3. Restart Claude Desktop: Close and reopen the application.

4. Verify: Click the hammer icon (Tools) in the chat input. You should see the databricks-genie tools listed (e.g., start_conversation, create_message).

Now you can ask Claude questions like "What was our DBU consumption last month?" or "Who accessed the PII table yesterday?", and it will use the tools pro

... View full README on GitHub