nfl-mcp

MCP server for NFL data (2013–2025), powered by nflreadpy and DuckDB. Query play-by-play, rosters, injuries, stats, and more using natural language in Claude Code, VS Code, or Claude Desktop.

Ask Claude questions like:

• "Who had the best EPA per play in 2024?"

• "Show me Patrick Mahomes' completion % over expected by season"

• "Compare 4th quarter red zone efficiency for KC vs PHI in 2023"

• "Which defenses had the highest sack rate in 3rd & long situations?"

• "Who was on IR for the Eagles in Week 10, 2023?"

• "Show me snap count trends for the Chiefs receiving corps in 2024"

Quickstart

pip install nfl-mcp        # or: uvx nfl-mcp
nfl-mcp init               # configure, load data, and start the server

init walks you through setup and offers to start the server immediately when done. No database server to install. No credentials to manage. Data is stored locally in DuckDB.

Prerequisites

• Python 3.10+

• uv (recommended) or pip

Setup

1. Initialize

nfl-mcp init

The wizard will:

1. Configure the local DuckDB database path

2. Download the default NFL datasets (play-by-play, rosters, stats, injuries, and more)

3. Auto-configure your IDE (Claude Desktop and/or VS Code)

4. Offer to start the server immediately

Options:

--skip-ingest       Configure without loading data

2. Start the server

init offers to start the server for you. If you need to start it manually later:

nfl-mcp serve
nfl-mcp serve --port 9000
nfl-mcp serve --host 0.0.0.0

The server uses the MCP Streamable HTTP transport. Point any MCP client at http://<host>:<port>/mcp.

Note: The server must be running for your IDE to connect. Run nfl-mcp serve in a terminal and keep it open.

3. Verify

nfl-mcp doctor

Checks database connectivity, loaded data, and IDE configuration.

4. Manual client configuration (optional)

If you skipped IDE setup during init, or need to reconfigure:

nfl-mcp setup-client                    # auto-detect clients
nfl-mcp setup-client --client vscode    # VS Code only
nfl-mcp setup-client --client claude-desktop

Or configure manually. Add to .vscode/mcp.json (VS Code):

{
  "servers": {
    "nfl": {
      "url": "http://localhost:8000/mcp"
    }
  }
}

Add to ~/Library/Application Support/Claude/claude_desktop_config.json (Claude Desktop):

{
  "mcpServers": {
    "nfl": {
      "url": "http://localhost:8000/mcp"
    }
  }
}

CLI Reference

nfl-mcp init               Interactive setup wizard
nfl-mcp serve              Start the MCP server (Streamable HTTP, default port 8000)
nfl-mcp ingest             Load NFL data into the database
nfl-mcp setup-client       Configure IDE MCP clients
nfl-mcp doctor             Health check

Serve options

nfl-mcp serve
nfl-mcp serve --port 9000
nfl-mcp serve --host 0.0.0.0

Ingestion options

nfl-mcp ingest                          # default datasets, all available seasons
nfl-mcp ingest --dataset all            # every dataset
nfl-mcp ingest --dataset schedules      # one specific dataset
nfl-mcp ingest --dataset pbp --dataset injuries   # multiple datasets
nfl-mcp ingest --start 2020 --end 2024  # limit to a season range
nfl-mcp ingest --fresh                  # re-ingest even if already loaded
nfl-mcp ingest --list                   # show all available dataset names

Ingest is idempotent — re-running skips datasets and seasons already in the database.

Datasets

All data is sourced from nflverse via nflreadpy and stored locally in DuckDB.

nfl-mcp ingest --dataset all   # load everything
nfl-mcp ingest --list          # see all dataset names

MCP Tools

Key columns in plays

• epa — expected points added (the best single-play quality metric)

• wpa — win probability added

• posteam / defteam — offensive/defensive team abbreviations

• passer_player_name / rusher_player_name / receiver_player_name

• play_type — 'pass' | 'run' | 'field_goal' | 'punt' | 'kickoff' | ...

• desc — raw play description (use ILIKE for text search)

Local Development

git clone https://github.com/ebhattad/nfl-mcp
cd nfl-mcp
pip install -e ".[dev]"

nfl-mcp ingest --dataset all --start 2024 --end 2024
nfl-mcp serve   # server available at http://localhost:8000/mcp
pytest
pytest -m unit     # unit tests
pytest -m integration  # integration tests (requires loaded DB)

Troubleshooting

• nfl-mcp doctor is the fastest way to verify config, database, and client setup.

• If tools return database errors, run nfl-mcp ingest to ensure data is loaded.

• You can override the DB location with NFL_MCP_DB_PATH=/path/to/nflread.duckdb.

• Re-running nfl-mcp ingest is safe — it skips anything already loaded.

License

MIT