Databricks MCP Server

MCP server for executing SQL queries on Databricks.

Built with Databricks SDK for Python.

Forked from https://github.com/RafaelCartenet/mcp-databricks-server.

1. Features

• Execute SQL queries on Databricks
• List available schemas in a catalog
• List tables in a schema
• Describe table schemas
• Unity Catalog metadata exploration (catalogs, schemas, tables)
• Table lineage information (upstream/downstream tables, notebooks)
• Block dangerous SQL commands for safety. The following statements are blocked:
  • DROP
  • DELETE
  • TRUNCATE
  • ALTER
  • CREATE
  • INSERT
  • UPDATE
  • MERGE
  • GRANT
  • REVOKE

2. Setup

2.1. Install uv

See uv installation guide: https://docs.astral.sh/uv/getting-started/installation/.

2.2. Clone

git clone https://github.com/i9wa4/mcp-databricks-server

2.3. Configure Databricks Authentication

Create ~/.databrickscfg for authentication.

Example:

# Personal Access Token
[DEFAULT]
host = https://your-workspace.cloud.databricks.com
token = dapi_your_personal_access_token
warehouse_id = your_warehouse_id

or

# Service Principal (OAuth M2M)
[DEFAULT]
host = https://your-workspace.cloud.databricks.com
client_id = your_client_id
client_secret = your_client_secret
warehouse_id = your_warehouse_id

3. Running

Example MCP Server configuration:

{
  "mcpServers": {
    "databricks": {
      "type": "stdio",
      "command": "uv",
      "args": [
        "--directory",
        "/path/to/mcp-databricks-server",
        "run",
        "mcp-databricks-server"
      ],
      "env": {
        "DATABRICKS_CONFIG_PROFILE": "DEFAULT"
      },
      "disabled": false
    }
  }
}

DATABRICKS_CONFIG_PROFILE environment variable is optional. If not set, the DEFAULT profile from ~/.databrickscfg is used.

4. Contributing

See CONTRIBUTING.md.