AnalyticDB PostgreSQL MCP Server provides a universal interface between AI agents and AnalyticDB PostgreSQL databases, enabling seamless communication to retrieve database metadata and execute SQL operations. It can be installed from source for development or from PyPI for production use. The server supports two transport modes: stdio (default) for MCP client integration and streamable HTTP for REST-based access, testing, or debugging. To connect to AnalyticDB PostgreSQL, you must configure environment variables such as ADBPG_HOST, ADBPG_PORT, ADBPG_USER, ADBPG_PASSWORD, and ADBPG_DATABASE, and to initialize graphRAG and LLM memory services, API_KEY, BASE_URL, LLM_MODEL, and EMBEDDING_MODEL. The MCP server exposes a set of built-in tools for executing queries and managing knowledge graphs and memories, including SELECT, DML, and DDL SQL execution, table statistics collection, and query explanation. Graphrag integration supports uploading text content to build knowledge graphs and querying them, while LLM memory endpoints enable storing and retrieving long-term memories. The README also provides MCP client integration examples for stdio and HTTP transports and related environment variables.