Velocirag

Velocirag is a lightning-fast, local-first Model Context Protocol (MCP) server specialized in retrieval-augmented generation (RAG) for AI assistants and agentic workflows. It exposes search, document indexing, knowledge graph traversal, and system diagnostics tools to MCP-compatible clients like Claude, Cursor, and Windsurf. Its unique 4-layer search architecture fuses vector similarity, BM25 keyword matching, knowledge graph traversal, and metadata filtering, all accelerated by ONNX Runtime for CPU-only inference. Velocirag is ideal for developers, AI researchers, data engineers, and organizations needing sub-200ms RAG search, local document intelligence, or robust agent integration without GPU dependencies or API keys.