Pybaseball MCP Server

This is a FastAPI-based MCP server that exposes MLB and Fangraphs baseball data via the pybaseballstats library. The pybaseballstats package relies on common scientific Python libraries such as pandas and numpy.

Features

• /player?name=... — Get player Statcast data by name (optionally filter by date range)
• /team_stats?team=...&year=...&type=batting|pitching — Get team batting or pitching stats for a given year
• /leaderboard?stat=...&season=...&league=...&type=batting|pitching — Get MLB leaderboard for a stat, season, and league

Setup

pip install -r requirements.txt
uvicorn main:app --reload

Example Usage

• Player stats:
  • GET /player?name=Mike Trout
  • GET /player?name=Shohei Ohtani&start_date=2023-04-01&end_date=2023-10-01
• Team stats:
  • GET /team_stats?team=Yankees&year=2023&type=batting
  • GET /team_stats?team=Dodgers&year=2022&type=pitching
• Leaderboards:
  • GET /leaderboard?stat=HR&season=2023&league=AL&type=batting
  • GET /leaderboard?stat=ERA&season=2023&league=NL&type=pitching

API Documentation

One up and running, interactive API docs are available at http://127.0.0.1:8000/docs.

Publishing

To expose an MCP STDIO interface instead of HTTP, first start the FastAPI server and then run the wrapper:

uvicorn main:app --host 0.0.0.0 --port 8000 &
python mcp_stdio_wrapper.py

This project is ready for deployment on Smithery or any other MCP-compatible platform. It uses the pybaseballstats library and its dependencies, including pandas and numpy.

Listing available tools

The server supports the MCP tool-discovery protocol. After starting the server, you can list available tools using either HTTP or the STDIO wrapper.

HTTP

curl -X POST http://localhost:8000/tools/list

STDIO

echo '{"jsonrpc": "2.0", "method": "tools/list", "id": 1}' | python mcp_stdio_wrapper.py

Powered by pybaseballstats and FastAPI