# PyRAG Core

An open-source, Python-first, API-driven Retrieval-Augmented Generation platform with FastAPI endpoints, LangGraph-native agents, a pluggable vector store, and support for several LLM providers.

PyRAG Core is a production-oriented, open-source RAG platform built around a FastAPI and Pydantic v2 API, LangGraph and LangChain agents, Celery and Redis for queuing, PostgreSQL with SQLAlchemy for relational data, and a pluggable vector store defaulting to Qdrant but also supporting Weaviate, Milvus, pgvector, and Elasticsearch. It supports multiple LLM providers, including OpenAI, Anthropic, Gemini, OpenRouter, Ollama, and vLLM, and uses MinIO for object storage and Langfuse for observability.

Local development runs entirely through Docker Compose and a Makefile: cloning the repository, syncing dependencies with `uv`, configuring environment variables, starting infrastructure containers, running database migrations, and starting the API and an optional Streamlit developer UI that offers a lightweight playground for the parsing, cleaning, and chunking pipeline stages without needing the full database or object storage running.

The project documents its architecture, data model, security model, and known limitations in a separate architecture document.

Website: https://github.com/SheikhSarvar/pyrag-core