Sports AI in Your Terminal & Group Chat
A CLI and bot scaffold that connects any LLM to live sports data via sports-skills. Ask questions in plain English, get real answers from real data.
curl -fsSL https://sportsclaw.gg/install.sh | bash
███████╗██████╗ ██████╗ ██████╗ ████████╗███████╗ ██████╗██╗ █████╗ ██╗ ██╗ ██╔════╝██╔══██╗██╔═══██╗██╔══██╗╚══██╔══╝██╔════╝██╔════╝██║ ██╔══██╗██║ ██║ ███████╗██████╔╝██║ ██║██████╔╝ ██║ ███████╗██║ ██║ ███████║██║ █╗ ██║ ╚════██║██╔═══╝ ██║ ██║██╔══██╗ ██║ ╚════██║██║ ██║ ██╔══██║██║███╗██║ ███████║██║ ╚██████╔╝██║ ██║ ██║ ███████║╚██████╗███████╗██║ ██║╚███╔███╔╝ ╚══════╝╚═╝ ╚═════╝ ╚═╝ ╚═╝ ╚═╝ ╚══════╝ ╚═════╝╚══════╝╚═╝ ╚═╝ ╚══╝╚══╝
How It Works
sportsclaw is a TypeScript CLI that connects LLMs to sports-skills, a Python package with 14 deterministic sports data connectors.
sports-skills
The deterministic data layer.
- Zero API keys required
- ESPN, FastF1, Polymarket
- Normalizes complex JSON
- Exposes JSON schemas to LLM
sportsclaw
The CLI & bot scaffold.
- Anthropic, OpenAI, Google
- Interactive setup & config
- Persistent memory
- Built-in Discord/Telegram bots
Readable Codebase
~2,300 lines of TypeScript built on the Vercel AI SDK. The core agent loop is ~80 lines. Easy to fork, extend, or learn from.
Zero API Keys for Data
You need an Anthropic or OpenAI key for the reasoning model, but the Python data layer requires absolutely zero API keys or paid subscriptions. It just works.
Interactive CLI
Setup is entirely interactive. Run it once, paste your API key, and watch the reasoning spinner solve your queries in real-time.
Discord & Telegram Bots
Want it in your group chat? The CLI includes built-in listeners.
sportsclaw listen discord
sportsclaw listen telegram