ApeRAG

Why we included this project

ApeRAG is a self-hostable retrieval platform that combines graph-based retrieval, vector search, and full-text search in a single service with one API and web UI, so a team gets a knowledge base without stitching together several RAG components. The graph layer is the reason to look twice: documents become a knowledge graph you can open and inspect, which means when a generated answer looks wrong you can trace the retrieval path instead of trusting a black-box vector store. It also handles multimodal documents and ships built-in agents, and an MCP endpoint lets an external AI assistant query the knowledge base over a standard protocol. Deployment runs from Docker Compose on a laptop to Kubernetes for production, so you can start small and grow without switching platforms.