The sports MCP server.
Real stats and pre-computed analysis in Claude, Cursor, ChatGPT, or any MCP client. Eight tools shaped for LLM context, backed by live ingestion. Sign in once. No API keys.
One URL, one sign-in. No keys to paste.
Point your client at mcp.statshawk.ai/mcp. One line in Claude, Cursor, ChatGPT, or any MCP-aware client.
A browser window, one sign-in. We issue and rotate the token. There is no API key to paste into a config file.
Rosters, box scores, standings, prop analysis. Your assistant calls the tools and answers from real numbers.
A catalog your model can actually hold in context.
One get_player_props that takes a league, not eight league-prefixed copies. Smaller catalog, faster tool selection, fewer wasted tokens.
Find games by league, date, team, or status.
Per-player stat lines, normalized per sport.
Ranked standings with W/L/T, win percentage, and points for and against.
Who is on the team right now, each with a person_id that chains into props.
Global fuzzy player search; returns a person_id for chaining.
Prop cards: season, recent, and home/away averages, with hit rates vs your line.
Pregame MLB matchups with pitching and lineup context.
MLB plate appearances and Statcast pitches, filterable per player.
Grounded answers, not vibes.
Who's on the Yankees roster, and how has their shortstop been hitting?
Here's the full roster by position. For the shortstop I pulled his prop card: season, recent, and home/away averages, and how often he clears 1.5 hits. Want the same for anyone else?
Build a fantasy assistant on real stats.
Draft prep, waiver targets, start/sit calls: fantasy decisions are stats questions, and your assistant can finally answer them with current data. Recent lines, matchup context, and prop-style hit rates, a tool call away.
You bring the league logic: your scoring rules, your roster. StatsHawk brings the numbers underneath. We serve league-wide stats and analysis, not your fantasy platform's account data.
> Who should I start at SS this week, Volpe or Winn?
> Which waiver-wire outfielders have the best L14 lines?
> Is anyone on my roster facing a top-5 strikeout staff?
5,000 units a month, every stat endpoint, hosted MCP included.
Sign up, connect your client, ask your first question. Analysis tools like player prop cards come with any paid plan.
Smithery ↗
Glama ↗Questions, answered.
What is an MCP server?
MCP (Model Context Protocol) is an open standard that lets AI assistants like Claude, Cursor, and ChatGPT call external tools. An MCP server exposes those tools. StatsHawk's hosted MCP server gives your assistant eight tools for real sports stats and analysis, so it answers from live data instead of guessing from training data.
Which AI clients does StatsHawk work with?
Any MCP-aware client. Claude (Desktop and Code), Cursor, and ChatGPT are the most common. If your client speaks MCP over streamable HTTP, it can connect to mcp.statshawk.ai/mcp.
Do I need an API key?
No. The hosted MCP server uses OAuth: your client opens a sign-in window once, and StatsHawk issues and rotates the token. API keys exist separately for the REST API, and both share one account and one usage meter.
What sports and data are covered?
Normalized stats across NBA, MLB, NFL, and NHL: box scores, standings, rosters, player search, play-by-play with Statcast for MLB, and pre-computed player prop analysis.
Is there a free tier?
Yes. Free includes 5,000 weighted units per month and every standard stat endpoint over hosted MCP. Analysis tools like player prop cards are part of every paid plan.