Hybrid Memory MCP Server

Example Use Cases

• 1

  Persistent memory with both keyword and semantic search capabilities.

• 2

  That require scalable, stateless memory storage backed by serverless PostgreSQL.

• 3

  That benefit from hybrid search combining full-text and vector similarity for improved recall.

Description

This MCP server combines traditional keyword-based full-text search with modern vector similarity search to enable efficient and scalable memory retrieval for AI agents. It supports memory operations like write, search, get, and delete, and is designed for production deployment with serverless Neon PostgreSQL and the FastMCP framework. The hybrid architecture enhances memory recall by leveraging both semantic embeddings and keyword matching, making it suitable for complex agent memory management.