LLM Chess

LLM Chess is a lightweight browser-based chess experience where you play against a large-language-model-powered opponent using your own OpenAI API key.

How it works

• The UI (index.html, static/js/game.js, static/css/style.css) renders a chessboard via Chessboard.js + Chess.js.
• When it's the AI's turn, the frontend posts directly to https://api.openai.com/v1/chat/completions with the stored OpenAI key so the AI can reason about the position and return a JSON move.
• Illegal-move detection and a simple overlay enforce a clean experience, while API keys live only in the user's browser localStorage.
• A legacy Flask server (app.py) still exists but is currently dormant and not used by the UI; you can remove it if you want a pure static deployment.

Getting started

1. Install dependencies and run a local dev server (optional):

   python -m venv venv
   source venv/bin/activate
   pip install -r requirements.txt
   python app.py

   The Flask app serves the static pages on http://localhost:5001, but the UI now calls OpenAI directly.
2. Alternatively, open index.html in a static hosting environment (GitHub Pages, Netlify, etc.).
3. Click the settings icon, paste your OpenAI API key, and start a new game. The key is stored only in your browser.

Notes

• Only models starting with gpt are listed; feel free to extend the dropdown if you plan to support other providers again.
• Because the browser makes requests straight to OpenAI, make sure your key stays private (don’t paste it in shared browsers).
• The project keeps the Flask backend around in case you want to reintroduce a proxy, but it’s not required for the current experience.