Skip to main content
SportsClaw is an agent CLI, built by Machina Sports, that connects any LLM (Anthropic, OpenAI, Gemini) to sports data — via the open-source Sports Skills package for public data, and via MCP for a connected Machina project’s live data. Full product docs live at sportsclaw.gg; this page only covers the Machina side of the integration.

​
Install

​
Connecting SportsClaw to a Machina project

The two CLIs are designed to pair directly. On the Machina side, machina connect resolves your project’s MCP endpoint and pairs it with an auth token in one step:
That prints a URL, transport (sse), and token — which SportsClaw’s own mcp add command consumes:
Server details (URL, name) are saved to ~/.sportsclaw/mcp.json; the token itself is kept separately in ~/.sportsclaw/.env, not written into that config file. Once added, SportsClaw discovers the project’s exposed tools over MCP the same way any other MCP client would — see MCP for what “exposed tools” means on the Machina side (that’s MCP Builder territory).
  • Connecting External Agents — the same MCP connection flow for other clients (Claude Code, Cursor, Codex).
  • connect, mcp — the Machina-side CLI commands used above.
  • Sports Skills — the open-data layer SportsClaw runs on before/alongside a Machina connection.