RAG system with vector storage and AI integration.
Obelisk is a RAG (Retrieval-Augmented Generation) system designed to provide context-aware AI interactions. It features a robust architecture for document processing, vector embedding, and AI-enhanced responses through integration with Ollama and OpenAI-compatible endpoints.
- RAG Pipeline: Complete document processing, embedding, and retrieval pipeline
- Vector Storage: Integration with ChromaDB and planned Milvus support
- OpenAI-Compatible API: Drop-in replacement for OpenAI API clients
- Ollama Integration: Local LLM support for both generation and embeddings
- Document Monitoring: Real-time updates when documents change
- Docker Support: Containerized deployment for all components
- Python Src Layout: Modern Python project structure with clean separation of concerns
- Type Annotations: Comprehensive type hints throughout the codebase
- Python 3.12+
- Poetry
- Docker (optional, for containerized usage)
# Clone the repository
git clone https://github.com/usrbinkat/obelisk.git
cd obelisk
# Install dependencies
poetry installObelisk provides a CLI for interacting with the RAG system:
# Show version
poetry run obelisk --version
# Show help
poetry run obelisk --help
# Index documents in the vault
poetry run obelisk rag index --vault /path/to/vault
# Query the RAG system
poetry run obelisk rag query "What is RAG?"
# Show system statistics
poetry run obelisk rag stats
# Start the RAG API server with OpenAI-compatible endpoint
poetry run obelisk rag serve --host 0.0.0.0 --port 8000 --watch# Run the full stack with Docker Compose
docker compose -f deployments/docker/compose/dev.yaml up
# Run only the RAG service
docker compose -f deployments/docker/compose/dev.yaml up obelisk-rag
# Run with Ollama integration
docker compose -f deployments/docker/compose/dev.yaml up ollama obelisk-ragPlease see CONTRIBUTING.md for details on how to contribute to this project.