Databricks MCP Proxy Server

Proxies Databricks MCP tools for MCP clients with OAuth authentication and remote tool discovery.

other

GitHub Stars

other

Language

6 months ago

First Indexed

3 months ago

Catalog Refreshed

Documentation & install

Readme and setup notes from the catalogue, plus a client-ready config you can copy for your MCP host.

Installation

Add the following to your MCP client configuration file.

Configuration

View docs

{
  "mcpServers": {
    "smaheshwari-ux-databricks-mcp-proxy": {
      "command": "uv",
      "args": [
        "run",
        "databricks-mcp-proxy"
      ],
      "env": {
        "DATABRICKS_HOST": "https://dbc-XXXXX.cloud.databricks.com",
        "DATABRICKS_APP_URL": "https://your-mcp-app.databricksapps.com"
      }
    }
  }
}

You can run a Databricks MCP Proxy to authenticate with Databricks and expose remote MCP tools to your MCP clients. This proxy acts as a bridge, allowing your local MCP workflow to discover and invoke tools hosted in your Databricks environment through a secure, authenticated channel.

How to use

You start the proxy locally and then connect your MCP client to the running proxy. Use the available tools to authenticate with Databricks, discover remote MCP tools, and invoke them from your MCP client. The workflow is designed to be practical: launch the proxy, perform authentication, let tools register, and then call any remote tool by name.

Practical usage pattern:

Start the proxy from your development environment.
Initiate the authentication flow when prompted by the client tool.
After authentication completes, browse or query the list of available remote tools.
Invoke a tool by its name to perform the desired action through the proxy.

How to install

uv sync

Configure the proxy with your Databricks details. Create or edit the configuration to include your Databricks host and the MCP app URL.

yaml
env:
  - name: DATABRICKS_HOST
    value: "https://dbc-XXXXX.cloud.databricks.com"

  - name: DATABRICKS_APP_URL
    value: "https://your-mcp-app.databricksapps.com"

Additional sections

Configuration details are provided to ensure the proxy can authenticate against Databricks and expose its MCP tools to your client. Review the environment variable values and verify the application URL points to your MCP app in Databricks.

Local test flow and desktop integration are supported. To test locally, run the proxy with the dedicated CLI command and then proceed with the authentication and tool discovery steps.

uv run databricks-mcp-proxy

json
{
  "mcpServers": {
    "databricks": {
      "command": "uv",
      "args": [
        "run",
        "databricks-mcp-proxy"
      ]
    }
  }
}

Claude Desktop can be configured to launch the proxy automatically. Add the following to your Claude desktop configuration to run the proxy from a specified directory.

{
  "mcpServers": {
    "databricks": {
      "command": "uv",
      "args": [
        "--directory", "/path/to/databricks-mcp-proxy",
        "run",
        "databricks-mcp-proxy"
      ]
    }
  }
}

Available tools

authenticate

Start OAuth flow to authenticate with Databricks and authorize the proxy to access Databricks resources.

list_databricks_tools

List available remote MCP tools exposed by the Databricks environment.

call_databricks_tool

Invoke a specific remote tool by its name to perform actions through the proxy.

Built by

VeilStrat

AI signals for GTM teams