How to Connect Databricks to Cursor via MCP

💡 No install? Use cloud.anythingmcp.com directly. Sign in, click Connectors → Databricks, paste your credentials, mint an MCP API key and you're done. No Docker, no git clone, no local server.

Databricks + Cursor

Let your AI work with your Databricks workspace: run SQL on a warehouse, explore Unity Catalog catalogs, schemas and tables, check clusters, warehouses, jobs and runs, read notebooks, and start or stop compute or trigger jobs after your confirmation. With a personal access token.

Prerequisites

See the full setup instructions baked into the connector (visible in the in-app store when you select the connector). The required environment variables for this connector are:

DATABRICKS_HOST, DATABRICKS_TOKEN

Step 1 — Get credentials

1. Open your Databricks workspace and copy its host from the address bar into DATABRICKS_HOST, without https://, for example dbc-a1b2c3d4-e5f6.cloud.databricks.com (AWS), adb-1234567890123456.7.azuredatabricks.net (Azure) or 1234567890123456.7.gcp.databricks.com (Google Cloud).
2. Click your user name at the top right, then Settings → Developer → Access tokens → Manage → Generate new token. Give it a comment and a lifetime and copy it into DATABRICKS_TOKEN: Databricks shows it only once. Your workspace admin must allow personal access tokens.
3. For least privilege, create the token for a service principal that has only the grants the AI needs: USE CATALOG, USE SCHEMA and SELECT on the data, and CAN USE on one SQL warehouse. The connector acts with the rights of the token's user.
4. Install the connector and run databricks_get_current_user to see who the AI acts as.

Compute costs money, so the AI asks first: Databricks bills compute in DBUs. A query on a stopped SQL warehouse starts it automatically, and a running warehouse or cluster costs money until it stops by itself. Before it runs SQL, the AI checks the warehouse with databricks_list_warehouses; if it is not running, it tells you the query will start it, and how big it is, and waits for your yes. A serverless warehouse starts in seconds.

SQL on a warehouse: databricks_execute_sql runs one statement and returns up to 200 rows by default (at most 10,000). For anything that changes data, tables or permissions (INSERT, UPDATE, DELETE, MERGE, CREATE, ALTER, DROP, GRANT) the AI shows you the exact statement and waits for an explicit yes; the token's Unity Catalog grants decide what is allowed at all. Long statements keep running in the background: databricks_get_statement follows them and databricks_get_statement_chunk fetches more rows.

Unity Catalog: databricks_list_catalogs, databricks_list_schemas, databricks_list_tables and databricks_get_table show catalogs, schemas, tables, views and their columns, so the AI writes SQL against the right names.

Every other change asks first too: databricks_start_cluster, databricks_terminate_cluster, databricks_start_warehouse, databricks_stop_warehouse, databricks_run_job, databricks_cancel_job_run and databricks_cancel_statement. Terminating a cluster or stopping a warehouse ends the queries and notebooks running on it. This connector deletes nothing.

Jobs: databricks_list_jobs and databricks_get_job show jobs, schedules, parameters and tasks; databricks_run_job starts a run now; databricks_list_job_runs, databricks_get_job_run and databricks_get_run_output show runs, task states, errors and notebook output (kept for 60 days).

Notebooks and files: databricks_list_workspace browses folders and notebooks, databricks_export_notebook returns the source code of a small notebook, databricks_list_volume_files lists a Unity Catalog volume and databricks_list_dbfs lists legacy DBFS paths. The Spark configuration, environment variables and init scripts of clusters are left out because they can contain secrets.

Common errors: 401, or 403 Invalid access token: the token is wrong, expired or revoked, or DATABRICKS_HOST points at another workspace. 403 PERMISSION_DENIED: the user lacks a grant, such as USE CATALOG, SELECT, CAN USE on the warehouse or CAN MANAGE RUN on the job. 404 RESOURCE_DOES_NOT_EXIST: wrong id or name, or Unity Catalog is not enabled for the workspace. 429: too many requests; wait and retry.

The connector does not create, edit or delete clusters, warehouses, jobs, tables or files, does not upload files, and does not cover Delta Live Tables pipelines, MLflow, Model Serving, Genie or Vector Search.

Step 2 — Install the adapter

curl -fsSL https://raw.githubusercontent.com/HelpCode-ai/anythingmcp/main/docker-compose.quickstart.yml -o docker-compose.yml
printf 'JWT_SECRET=%s\nENCRYPTION_KEY=%s\n' "$(openssl rand -hex 32)" "$(openssl rand -hex 32)" > .env
docker compose up -d

Step 3 — Add the connector in Cursor

Your server URL: in AnythingMCP, open MCP Servers → the server this connector is on, and copy its URL (https://cloud.anythingmcp.com/mcp/…). Use it wherever this guide shows YOUR_SERVER_ID.

Cursor reads MCP servers from ~/.cursor/mcp.json. Add this entry:

{
  "mcpServers": {
    "anythingmcp": {
      "url": "https://cloud.anythingmcp.com/mcp/YOUR_SERVER_ID",
      "headers": { "X-API-Key": "YOUR_MCP_API_KEY" }
    }
  }
}

1. Create an MCP API key in AnythingMCP under MCP Servers → your server → API keys, and paste it in place of YOUR_MCP_API_KEY.
2. Save the file and restart Cursor.
3. Open Cursor → Settings → MCP and check that anythingmcp is listed and connected.
4. Start chatting: the agent can now call the Databricks tools.

Available tools

Example prompts

• "Which SQL warehouses do we have, and which ones are running right now?"
• "Show me the tables in main.sales and the columns of the orders table."
• "What was our revenue by country last month? Use the serverless warehouse, and tell me first if it has to start."
• "Did last night's nightly_etl job succeed? If a task failed, show me its error."
• "Run the weekly_report job now, after you tell me what it does and which cluster it uses."

FAQ

Does Cursor support custom MCP servers on the free tier? Yes, MCP is available on Hobby, Pro and Business plans.

Next steps

• Connector landing page
• Connect to Claude
• Connect to ChatGPT