KBDB

A PostgreSQL-backed RAG (Retrieval-Augmented Generation) server that provides semantic search across document embeddings using pgvector for vector similarity operations. Built with FastMCP and OpenAI's embedding API, it supports multiple search modalities including question-answer pairs, semantic similarity, style-based clustering, and code similarity search, each optimized with different embedding prefixes and distance metrics (cosine, inner product, L2). The system stores documents in chunks with pre-computed embeddings and exposes specialized search tools for different use cases, making it valuable for knowledge base applications, documentation search, and content discovery workflows where different types of semantic matching are required.