{
  "markdown": "# Chess Coach AI\n\nAI-powered chess coaching system built as a **Claude Code plugin** with a multi-agent cowork architecture.\n\n## Features\n\n- **Intel Agent** — Player profiling, Lichess API integration, Opening Explorer analysis, opponent scouting\n- **GM Agent** — Stockfish/Maia analysis, error classification by cognitive origin, personalized training roadmaps, auto-analysis with training insights\n- **Mind Agent** — Tilt detection, time pattern analysis, psychological profiling, session readiness assessment\n- **Biohack Agent** — Evidence-based nutrition protocols, supplementation, sleep optimization, pre-game routines\n\n## Getting Started\n\n### Prerequisites\n\n```bash\n# Required\nbrew install stockfish          # Stockfish 18+\npython3 --version               # Python 3.12+\n```\n\n### Option A: Install via Marketplace (recommended)\n\nThe easiest way to install. In any Claude Code session:\n\n```bash\n# 1. Add the marketplace\n/plugin marketplace add datoga/datoga-plugins\n\n# 2. Install the plugin\n/plugin install chess-coach-ai@datoga-plugins\n```\n\nThat's it. The coach skill is now available in all your sessions.\n\n### Option B: Install from GitHub\n\n```bash\nclaude plugin install github:datoga/chess-coach-ai\n```\n\n### Option C: Install from local clone\n\n```bash\ngit clone https://github.com/datoga/chess-coach-ai.git\ncd chess-coach-ai\npip install -r requirements.txt\nclaude --plugin-dir ./chess-coach-ai\n```\n\n### Verify Installation\n\nOnce the plugin is loaded, run the setup wizard:\n\n```\n/chess-coach-ai:setup\n```\n\nThis checks all prerequisites (Stockfish, Python deps, opening database) and guides you through fixing any issues.\n\n### Available via the `/chess-coach-ai:coach` skill\n\nOnce loaded, the coach skill is available in any Claude Code session. Example commands:\n\n- `Prepare my game against [lichess username]`\n- `Review this game: [paste PGN]`\n- `Create a training plan`\n- `Intel on [lichess username]`\n- `Save this game: [paste PGN]`\n- `Import my games from lichess`\n- `Starting a training session — slept 6 hours, energy 7/10`\n\n## Architecture\n\n```\nUser → /chess-coach-ai:coach (Coordinator)\n       ├→ Intel (Lichess API, Opening Explorer, chessdb.cn)\n       ├→ GM (Stockfish analysis, PGN analysis, training insights)\n       ├→ Mind (time patterns, tilt detection, resilience)\n       └→ Biohack (nutrition, sleep, supplements, protocols)\n       → Coordinator synthesizes → Unified response\n```\n\nThe coordinator dispatches agents via **cowork** (Claude Code agent teams). Agents communicate through JSON contracts defined in `data/schemas/`.\n\n## Storage\n\nGames are stored in the **Game Vault**:\n- **Primary:** Google Drive (requires OAuth2 setup)\n- **Fallback:** Local `vault/` directory\n\nSupports: manual PGN paste, Lichess URL import, bulk game download.\n\n## Testing\n\n```bash\n# Run all tests\npytest tests/ -v\n\n# Run specific tool tests\npytest tests/test_pgn_parser.py -v\npytest tests/test_dqm_calculator.py -v\n```\n\n## Evals\n\nTrigger evals (does the right agent activate?) and quality evals (is the output correct?) are in:\n- `skills/coach/eval-set.json` + `skills/coach/evals/evals.json`\n- `evals/{intel,gm,mind,biohack}/`\n\n## Project Structure\n\n```\n├── agents/           # Agent definitions (Intel, GM, Mind, Biohack)\n├── skills/coach/     # Coordinator skill\n├── tools/            # Python tools (Lichess client, PGN parser, etc.)\n├── data/\n│   ├── schemas/      # JSON Schema contracts between agents\n│   ├── openings/     # lichess-org/chess-openings dataset\n│   ├── supplements.json\n│   └── nutrition_protocols.json\n├── templates/        # Output templates (dossiers, reviews, roadmaps)\n├── evals/            # Agent-level trigger + quality evals\n├── hooks/            # Quality gate hooks\n├── vault/            # Local game storage (gitignored)\n└── tests/            # pytest test suite\n```\n\n## License\n\nMIT\n",
  "bytes": 3829,
  "sha": "2c931abdad5f3979ced3d0fd48bc2fd3a2e7380d4abcf79467edd36c6a778369",
  "repo_slug": "datoga/chess-coach-ai",
  "fonte": "repo",
  "truncated": false,
  "api": "https://agentalog.com/api/listings/plg_datoga_chess_coach_ai_chess_coach_ai_23a803d4/readme"
}