MCP Databricks Server

🚀 Databricks MCP Server

This is a Model Context Protocol (MCP) server designed to execute SQL queries on Databricks via the Statement Execution API. It can use the Databricks API to perform SQL requests and retrieve data. When used in proxy mode, it can successfully iterate through multiple requests to execute complex tasks. It works best in conjunction with Unity Catalog metadata.

🚀 Quick Start

The Databricks MCP Server enables you to execute SQL queries on Databricks, list available schemas in catalogs, list tables in schemas, and describe table structures.

✨ Features

• Execute SQL queries on Databricks.
• List available schemas in a catalog.
• List tables in a schema.
• Describe the structure of a table.

📦 Installation

System Requirements

• Python version 3.10 or higher.
• If you plan to install via uv, ensure you have installed it.

Installation Steps

1. Install the required dependencies:

pip install -r requirements.txt

Or use uv:

uv pip install -r requirements.txt

2. Set up environment variables:

Option 1: Use a .env file (recommended)

Create a .env file containing your Databricks credentials:

DATABRICKS_HOST=your-databricks-instance.cloud.databricks.com
DATABRICKS_TOKEN=your-databricks-access-token
DATABRICKS_SQL_WAREHOUSE_ID=your-sql-warehouse-id

Option 2: Set via the command line

export DATABRICKS_HOST="your-databricks-instance.cloud.databricks.com"
export DATABRICKS_TOKEN="your-databricks-access-token"
export DATABRICKS_SQL_WAREHOUSE_ID="your-sql-warehouse-id"

📚 Documentation

Permission Requirements

• SQL Warehouse Permissions: Ensure that the user has permissions to execute queries.
• Catalog and Schema Access: Configure access to specific catalogs, schemas, and tables according to your organization's security policies.
• Best Practices:
  • Rotate access tokens regularly.
  • Avoid hard - coding sensitive information into your code.

Running the Server

Stand - Alone

python dbapi.py

Integration with Cursor

In Agent Composer or other MCP clients, you can use the following tools:

execute_sql_query("SELECT * FROM my_schema.my_table LIMIT 10")
list_schemas("my_catalog")
list_tables("my_catalog.my_schema")
describe_table("my_catalog.my_schema.my_table")

Handling Long - Running Queries

The server is designed to handle long - running queries by polling the Databricks API until the query is completed or times out. The default timeout is 10 minutes (60 retries with a 10 - second interval), which can be adjusted in the dbapi.py file as needed.

🔧 Technical Details

Dependencies

• httpx: Used to send HTTP requests to the Databricks API.
• python - dotenv: Used to load environment variables from a .env file.
• mcp: Model Context Protocol library.
• asyncio: Used for asynchronous operations.