A local RAG-powered Model Context Protocol (MCP) server using Qdrant and FastEmbed to help AI coding agents navigate, understand, and optimize large codebases.