Databricks MCP Server

A Python MCP server to interact with Databricks — run SQL queries, manage clusters, trigger jobs, run notebooks, explore Unity Catalog metadata, manage secrets, and interact with DBFS.

Setup

1. Install dependencies

pip install -r requirements.txt

2. Configure credentials

Edit the .env file:

DATABRICKS_HOST=https://<your-workspace>.azuredatabricks.net
DATABRICKS_TOKEN=<your-access-token>
DATABRICKS_SQL_HTTP_PATH=/sql/1.0/warehouses/<your-warehouse-id>
DATABRICKS_WAREHOUSE_ID=<your-warehouse-id>

3. Run the server

python -m databricks_mcp.server

---

Available Tools (40+)

SQL

Tool	Description
sql_query	Run a SQL query (auto-wakes warehouse)
sql_explain	Show execution plan of a SQL query

Clusters

Tool	Description
clusters_list	List all clusters and their states
cluster_status	Get current state of a specific cluster
cluster_start	Start a cluster
cluster_stop	Stop a cluster
cluster_restart	Restart a running cluster
cluster_resize	Scale cluster workers up/down
cluster_wait_until_ready	Poll until cluster is RUNNING
cluster_events	Get recent events/errors for a cluster

Warehouses

Tool	Description
warehouses_list	List all SQL Warehouses
warehouse_status	Get detailed status of a warehouse
warehouse_start	Start a SQL Warehouse
warehouse_stop	Stop a SQL Warehouse
warehouse_resize	Change warehouse size

Jobs

Tool	Description
jobs_list	List all Databricks jobs
job_create	Create a new job with a notebook task
job_update	Update a job's name, notebook, or cluster
job_delete	Delete a job
job_trigger	Trigger a job run
job_run_status	Get status of a job run
job_cancel	Cancel a running job run
job_run_history	List recent runs of a job

Notebooks

Tool	Description
notebooks_list	List notebooks in a workspace path
workspace_list	List all objects in a workspace path
workspace_mkdir	Create a folder in workspace
notebook_run	Run a notebook (fire and forget)
notebook_run_and_wait	Run a notebook and wait for output
notebook_export	Export a notebook (HTML/SOURCE/JUPYTER/DBC)
notebook_import	Import/upload a notebook to workspace
notebook_delete	Delete a notebook or folder

Metadata (Unity Catalog)

Tool	Description
catalogs_list	List all catalogs
schemas_list	List schemas in a catalog
tables_list	List tables in a catalog.schema
table_schema	Get column schema of a table
table_preview	Preview top N rows of a table
table_create	Create a Delta table from DDL
table_drop	Drop a table
table_optimize	Run OPTIMIZE (+ optional ZORDER)
table_vacuum	Run VACUUM to clean up old files

Secrets

Tool	Description
secrets_list_scopes	List all secret scopes
secrets_list	List secret keys in a scope (no values)

DBFS

Tool	Description
dbfs_list	List files in a DBFS path
dbfs_delete	Delete a file or folder from DBFS
dbfs_upload	Upload a file to DBFS (base64 content)
dbfs_mkdir	Create a directory in DBFS

---

Integrate with Claude Desktop

Add this to your claude_desktop_config.json:

{
  "mcpServers": {
    "databricks": {
      "command": "python",
      "args": ["-m", "databricks_mcp.server"],
      "cwd": "/path/to/MCP"
    }
  }
}