mcp-memory

mcp-memory

3.5

If you are the rightful owner of mcp-memory and would like to certify it and/or have it hosted online, please leave a comment on the right or send an email to henry@mcphub.com.

MCP Memory is a server that enables MCP Clients to remember user information across conversations using vector search technology.

MCP Memory is a Model Context Protocol (MCP) Server designed to provide MCP Clients, such as Cursor, Claude, and Windsurf, with the capability to remember user information, including preferences and behaviors, across multiple conversations. It leverages advanced vector search technology to retrieve relevant memories based on semantic meaning rather than just keywords. The system is built using Cloudflare Workers, D1, Vectorize (RAG), Durable Objects, Workers AI, and Agents, ensuring efficient and secure memory storage and retrieval. MCP Memory is designed to be easily deployable on Cloudflare's infrastructure, offering a free tier for most users, and supports isolated memory spaces for individual users.

Features

  • Semantic Memory Retrieval: Uses vector search to find relevant memories based on meaning.
  • Cloudflare Integration: Built with Cloudflare Workers, D1, and Vectorize for efficient operation.
  • Durable Objects: Manages state and ensures consistency in memory operations.
  • Secure and Isolated: User memories are stored in isolated namespaces with secure communication.
  • Free Tier: Offers a free tier for normal usage levels, including 1,000 memories and ~28,000 queries per month.