F1 Analytics MCP Server

An MCP (Model Context Protocol) server providing 118+ Formula 1 analytics tools powered by FastF1 and the Ergast API. Connect it to Claude Desktop, Claude Code, or any MCP client to analyze races, compare drivers, explore telemetry, and simulate strategies using natural language.

Features

• Session & Driver Info — race results, standings, circuit details, driver metadata
• Lap & Sector Analysis — lap times, sector deltas, consistency scores, clean lap filtering
• Telemetry — speed/throttle/brake/gear/RPM traces, corner analysis, braking points
• Strategy Intelligence — tyre degradation, pit windows, undercut/overcut analysis, strategy simulation
• Race Position — overtakes, battles, gap evolution, position changes, lead changes
• Predictive AI — race winner prediction, overtake probability, tyre cliff prediction, safety car probability
• Visualizations — speed maps, race progression charts, tyre degradation plots, track dominance maps

Quick Start

pip install (recommended)

pip install f1-mcp

Install from source:

pip install git+https://github.com/Luffy610/f1-mcp.git

Docker

docker compose up -d

The MCP server will be available at http://localhost:8000/sse.

MCP Client Configuration

Claude Desktop / Claude Code (stdio — recommended)

{
  "mcpServers": {
    "f1": {
      "command": "f1-mcp"
    }
  }
}

SSE mode (remote / Docker)

{
  "mcpServers": {
    "f1": {
      "type": "sse",
      "url": "http://localhost:8000/sse"
    }
  }
}

To start the SSE server manually: python server.py

Tool Categories

Example Prompts

Analyze why Verstappen won the 2023 Bahrain GP. Break down his tyre strategy,
pit stops, race pace trend, and any overtakes he made.

Compare Verstappen and Leclerc through Turn 10 at the 2023 Bahrain GP.
Who brakes later, carries more apex speed, and gets on the throttle earlier?

Simulate how the 2023 Bahrain GP would have played out if Verstappen used
a two-stop strategy instead of his actual strategy.

See example.md for more detailed example prompts with the specific tools they trigger.

Contributing

We welcome contributions! The master branch is protected — please fork the repo, create a feature branch, and submit a pull request.

1. Fork the repository
2. Create your feature branch (git checkout -b feature/my-new-tool)
3. Make your changes and ensure tests pass (python -m pytest tests/ -v)
4. Submit a pull request

See CONTRIBUTING.md for detailed guidelines on project structure, adding tools, and code style.

Reporting Issues & Requesting Features

• Bug reports — Open an issue with steps to reproduce
• Feature requests — Open an issue describing your use case
• Questions & ideas — Start a thread in GitHub Discussions

Development

# Install in editable mode with dev dependencies
pip install -e ".[dev]"

# Run all tests
python -m pytest tests/ -v

# Run specific test group
python -m pytest tests/ -v -k "test_lap"

# Start SSE server for development
python server.py

Architecture

f1_mcp/                → Python package
  server.py            → Entry point, registers all tool modules
  config.py            → FastF1 cache configuration
  connectors/          → Data sources (FastF1Loader, ErgastClient)
  core/                → Caching layer + serialization utilities
  models/              → Pydantic schemas
  services/            → Business logic (14 service classes)
  tools/               → MCP tool registrations (14 modules, 1:1 with services)
tests/                 → End-to-end test suite (118 tests)

Environment Variables

License

MIT