Sleeper Fantasy Football MCP
MCP server that scores every projection against your league's actual rules, for draft, waiver, trade, and lineup advice.
Scope: Solo, independent project. Tools: Python, MCP, Sleeper API.
Overview
An MCP (Model Context Protocol) server that gives league accurate fantasy football advice for Sleeper leagues. Generic rankings assume generic scoring. This re-scores every projection as a dot product against a league's actual scoring settings, then converts it to VOR (value over replacement) using that league's actual roster shape, before generating draft, waiver, trade, or lineup recommendations.
Features
- Draft advice: VOR rankings, position tiers, keeper analysis, and draft planning.
- In season management: waiver targets, trade evaluation, optimal lineups, matchup previews, and player analysis.
- Advanced analytics: Monte Carlo simulations, consistency reports, strength of schedule, Vegas lines, and weather impact.
Architecture
Built in Python 3.10 or later as an MCP server, pulling from the public, unauthenticated Sleeper API and ESPN, with optional OpenWeather data for weather impact. The scoring engine re-derives value from each league's own settings rather than relying on generic, one-size-fits-all rankings.
Testing
The test suite runs ~113 invariant checks against the scoring and recommendation logic. Source is MIT licensed.
Source Code
The full server, fantasy football logic modules, and test suite are on GitHub.