A comprehensive MCP server for PostgreSQL database interaction, integrating semantic intelligence via embeddings and RAG (Retrieval-Augmented Generation). Version 1.2.0, released regularly. Key differentiators: built-in high-performance memory (4096D) with pgvector support for SOTA embeddings, segregated agent memory per user, autonomous navigation allowing natural language queries without SQL. Provides both a CLI MCP server for tools like table browsing, query execution, and schema extraction, and a TypeScript library for programmatic embedding and hybrid search. Requires Node.js >=20 <25 and a PostgreSQL instance with pgvector extension. Competes with other PostgreSQL MCP servers but focuses on agent-oriented memory and semantic capabilities.
npm install overmind-postgres-mcpNo compatibility data collected yet for this library.
Verified import paths — ran on the pinned version, not inferred.
Shows two ways: CLI MCP server setup via .mcp.json and .env, and programmatic embedding usage.
Upgrade Node.js to v20 or v21/v22 (but not v25). Use nvm or similar.
Check documentation for available model options and pass explicit model parameter.
Ensure OPEN_ROUTER_API_KEY is set in environment or .env file.
Run 'CREATE EXTENSION IF NOT EXISTS vector;' on your PostgreSQL instance.
Use npx to avoid global conflicts, or install locally.
Install locally: `npm install overmind-postgres-mcp`, then use correct import path.
Use named exports from subpaths, e.g., `import { embedText } from 'overmind-postgres-mcp/services/embeddings'`.Set POSTGRES_URL in .env or environment variables.
Start PostgreSQL service and verify connection string.