mcp_massive

The Massive.com MCP Server exposes four composable tools that collectively cover the Massive.com API surface: search_endpoints, get_endpoint_docs, call_api, and query_data. Use search_endpoints to discover available endpoints and built-in functions by natural language queries, then narrow results by scope to endpoints, functions, or all. Once you have an endpoint, use get_endpoint_docs with the docs URL from search_endpoints to understand required parameters. Call API calls with call_api to retrieve data from Massive.com REST endpoints, optionally storing results as in-memory DataFrames and applying post-processing functions. Finally, query_data lets you run SQL over stored DataFrames (SHOW TABLES, DESCRIBE, CREATE/CTEs, window functions, etc.) and apply additional post-processing. Built-in functions span Greeks (pricing and Greeks like bs_delta), Returns (simple_return, sharpe_ratio), and Technicals (sma, ema) that can be applied via the apply parameter on call_api or query_data. The integration enables data exploration by loading data, performing calculations, and querying results directly from an LLM workflow.

Prerequisites:

• Python 3.10+ installed on your system
• An active Massive.com API key
• Astral UV installed (v0.4.0+) for initial setup

1. Install the MCP server in your Python environment

uv tool install "mcp_massive @ git+https://github.com/massive-com/mcp_massive@v0.8.1"

2. Authenticate and run the server locally

MASSIVE_API_KEY=<your_api_key_here> uv run mcp_massive

3. (Optional) If you want to integrate with Claude Code tooling, install Claude Code and add the MCP server as shown in the Claude Code setup section of the README. You can also upgrade later with:

uv tool upgrade mcp_massive

4. Configuration for Claude Desktop (example) shows how to wire up the server with your config file. You can adapt the provided example to point to your local installation or to the path of your installed binary if you are using a different setup.

Prerequisites recap:

• Python 3.10 or newer
• MASSIVE_API_KEY environment variable (or equivalent in config)
• UV (Astral UV) installed for initial setup
• Network access to Massive.com API endpoints

Tips and caveats:

• The MCP server indexes Massive.com's endpoints at startup from the llms.txt file, so restart may be needed after API surface changes.
• The transport default is stdio; you can switch to streamable-http by setting MCP_TRANSPORT or using the appropriate CLI option.
• Ensure your MASSIVE_API_KEY is kept secure; do not expose it in shared config files.
• If migrating from uvx/uv run approaches, prefer installing the MCP server with uv tool install to download dependencies once for faster startup.
• For Claude integration, the MASSIVE_API_KEY should be provided in environment variables for the running MCP server used by Claude.
• You can adjust MASSIVE_MAX_TABLES and MASSIVE_MAX_ROWS to tune memory usage for your workload.