Databricks MCP Server

@JustTryAI

关于 Databricks MCP Server

A Model Completion Protocol (MCP) server for interacting with Databricks services

配置

使用下面的配置,将此服务器添加到你的 MCP 客户端。

{
  "mcpServers": {
    "databricks-mcp-server-justtryai": {
      "command": "uv",
      "args": [
        "venv"
      ]
    }
  }
}

工具

List all Databricks clusters

Create a new Databricks cluster

Terminate a Databricks cluster

Get information about a specific Databricks cluster

Start a terminated Databricks cluster

List all Databricks jobs

Run a Databricks job

List notebooks in a workspace directory

Export a notebook from the workspace

List files and directories in a DBFS path

Execute a SQL statement

概览

What is Databricks MCP Server?

The Databricks MCP Server is a Model Context Protocol (MCP) server that provides access to Databricks functionality via the MCP protocol. It allows LLM-powered tools to interact with Databricks clusters, jobs, notebooks, and more.

How to use Databricks MCP Server?

Install the server using the uv package manager (Python 3.10+ required), set the DATABRICKS_HOST and DATABRICKS_TOKEN environment variables for authentication, then run the provided start script (start_mcp_server.sh on Linux/Mac or start_mcp_server.ps1 on Windows). The server exposes Databricks operations as MCP tools.

Key features of Databricks MCP Server

MCP protocol support for LLM interaction with Databricks.
Databricks REST API integration for core operations.
Tool registration exposing Databricks functionality as MCP tools.
Async support built with asyncio for efficient operation.

Use cases of Databricks MCP Server

List and manage Databricks clusters (create, terminate, start, get info).
Run and list Databricks jobs.
Browse and export notebooks from the workspace.
List files and directories in a DBFS path.
Execute SQL statements on Databricks.

FAQ from Databricks MCP Server

What are the prerequisites for using Databricks MCP Server?

Python 3.10 or higher and the uv package manager are required. A Databricks workspace with a personal access token is also needed.

How do I authenticate with Databricks?

Set the environment variables DATABRICKS_HOST (your Databricks instance URL) and DATABRICKS_TOKEN (your personal access token). Alternatively, use a .env file.

What tools does the server expose?

The server exposes tools for clusters (list, create, terminate, get, start), jobs (list, run), notebooks (list, export), files (list in DBFS), and SQL (execute a statement).

Does it support running SQL queries?

Yes, the server includes an execute_sql tool that can execute a SQL statement.

How do I start the server after installation?

Run the wrapper script start_mcp_server.sh (Linux/Mac) or start_mcp_server.ps1 (Windows). The scripts are in the repository root and will start the MCP server ready for connections.

常见问题

What are the prerequisites for using Databricks MCP Server?

Python 3.10 or higher and the `uv` package manager are required. A Databricks workspace with a personal access token is also needed.

How do I authenticate with Databricks?

Set the environment variables `DATABRICKS_HOST` (your Databricks instance URL) and `DATABRICKS_TOKEN` (your personal access token). Alternatively, use a `.env` file.

What tools does the server expose?

The server exposes tools for clusters (list, create, terminate, get, start), jobs (list, run), notebooks (list, export), files (list in DBFS), and SQL (execute a statement).

Does it support running SQL queries?

Yes, the server includes an `execute_sql` tool that can execute a SQL statement.

How do I start the server after installation?

Run the wrapper script `start_mcp_server.sh` (Linux/Mac) or `start_mcp_server.ps1` (Windows). The scripts are in the repository root and will start the MCP server ready for connections.