Skip to main content
Glama
JustinUbes

rag-nfl-analyst

by JustinUbes
README.md
# rag-nfl-analyst

RAG system and MCP server for NFL fantasy football applications.

This repository is the retrieval and context-serving layer for the fantasy-football-ai project. It owns the NFL data-source code that was split out of the app repo, plus the first MCP server entrypoint for retrieval-style tools.

## What lives here

- NFL stats ingestion helpers from `nfl-data-py`
- MCP server entrypoint and tool surface
- Future retrieval, embedding, indexing, and citation plumbing

## Current layout

- `src/rag_nfl_analyst/data/nfl/stats.py` - weekly NFL stats and related data helpers
- `src/rag_nfl_analyst/mcp/server.py` - minimal MCP server with retrieval-oriented tools
- `docs/data-sources.md` - current source map and usage notes
- `specs/implementation-plan.md` - phased implementation plan for the retrieval stack
- `specs/open-questions.md` - remaining decisions and design questions

## Current state

The repo is intentionally thin right now:

- source integrations are in place
- an MCP server entrypoint exists
- retrieval/indexing/storage layers are still planned

## Getting started

```bash
uv sync
uv run rag-nfl-analyst
```

## Notes

- This repo is the right place for retrieval-specific code, not the final user-facing CLI or prompt formatting.
- The app repo owns Sleeper and user/platform state; this repo only receives normalized context.