An intelligent agentic RAG (Retrieval-Augmented Generation) system for chess knowledge, running entirely locally using Ollama.

The following sections are just an example of what I would like to implement. Everything is a working process.

• Agentic Intelligence: ReAct pattern with multi-step reasoning
• Knowledge Base: Chess theory, openings, historical games, strategies
• Tool Integration: ELO fetcher, PGN parser, game search, Stockfish (optional)
• Local First: Runs 100% locally with Ollama
• REST API: FastAPI with WebSocket support for streaming

• LLM Backend: Ollama (DeepSeek, Qwen)
• RAG Framework: LlamaIndex
• Vector DB: ChromaDB
• API: FastAPI
• Tools: python-chess, requests, stockfish (optional)

• Python 3.11+
• uv package manager
• Ollama

1. Clone the repository:

   git clone https://github.com/silvano315/chess-agentic-rag.git
   cd chess-agentic-rag

2. Install dependencies:

   uv sync

3. Setup Ollama:

   bash scripts/setup_ollama.sh

4. Configure environment:

   cp .env.example .env
   # Edit .env with your settings

5. Run the API:

   uv run uvicorn src.api.main:app --reload

See docs/ARCHITECTURE.md for detailed architecture.

• M0: Setup environment
• M1: Data pipeline
• M2: Vector store + Simple RAG
• M3: Tool implementation
• M4: Agentic orchestrator
• M5: Memory system
• M6: FastAPI interface
• M7: Advanced features

See docs/milestones/ for detailed milestone documentation.

# Run this command to install dev dependecies
uv sync --extra dev

# Run all tests
uv run pytest

# Run unit tests only
uv run pytest tests/unit

# Run with coverage
uv run pytest --cov=src --cov-report=html

MIT License - see LICENSE file.