MindsDB is an MCP server that enables your MCP applications to answer questions over large-scale federated data spanning databases, data warehouses, and SaaS applications.

​
Start MindsDB as an MCP Server

Follow the steps below to use MindsDB as an MCP server.
  1. Install MindsDB locally via Docker or Docker Desktop.
  2. Connect your data source and/or upload files to MindsDB in order to ask questions over your data.
  3. Start MindsDB MCP server, either with or without authentication.
    • Start MindsDB MCP server without authentication to connect it to Cursor. 
      docker run --name mindsdb_container -e MINDSDB_APIS=http,mysql,mcp,a2a -p 47334:47334 -p 47335:47335 -p 47337:47337 -p 47338:47338 mindsdb/mindsdb
    • Start MindsDB MCP server with authentication to connect it to OpenAI or Anthropic. 
      docker run --name mindsdb_container -e MINDSDB_MCP_ACCESS_TOKEN=abc123 -e MINDSDB_APIS=http,mysql,mcp,a2a -p 47334:47334 -p 47335:47335 -p 47337:47337 -p 47338:47338 mindsdb/mindsdb
    The following environment variables are set in the Docker container:
    • MINDSDB_APIS defines the MindsDB’s APIs to be started.
    • MINDSDB_MCP_ACCESS_TOKEN defines the MCP access token used to access MindsDB’s MCP Server from MCP Clients.
    Alternatively, you can use the MINDSDB_APIS environment variable to define the list of APIs to be started.
    export MINDSDB_APIS='mcp'
    If you run MindsDB via PyPI, you can define APIs as below.
    python -m mindsdb --api=mcp

​
MCP Tools

MindsDB MCP API exposes a set of tools that enable users to interact with their data and extract valuable insights. 1. List Databases The list_databases tool lists all data sources connected to MindsDB. 2. Query The query tool executes queries on the federated data to extract data relevant to answering a given question.