Back to projects
A local, free PostgreSQL internals assistant — semantic search over PG source code integrated into Zed and VSCode via the Model Context Protocol (MCP). Ask natural language questions about PostgreSQL internals without any cloud API or internet dependency.
PostgreSQLPythonMCPpgvectorOllamaRAG
Features
- ask_pg: RAG combining semantic search with a local LLM grounded in actual PG source
- search_pg: Direct semantic search returning top-matching source code chunks
- search_symbol: cscope-based lookups for callers, definitions and references
- Entirely local — no external API calls or data transmission
- pgvector HNSW index over 33,734 PostgreSQL source chunks with nomic-embed-text embeddings
- Integrates with Zed and VSCode Agent Panels via FastMCP