Memory
CreatedPuliczek
Enables MCP clients to remember information about users across conversations, leveraging vector search for relevant memory retrieval.
About
Memory is a server that allows MCP clients like Cursor, Claude, and Windsurf to maintain user context across multiple conversations. By utilizing vector search technology, it efficiently retrieves relevant memories based on meaning rather than keywords. Built with Cloudflare Workers, D1, Vectorize, Durable Objects, Workers AI, and Agents, Memory offers a cost-effective solution (often free for typical usage) for adding persistent memory capabilities to MCP-based applications.
Key Features
- Stores text and embeddings in Cloudflare Vectorize and D1
- Provides vector similarity search for contextual memory retrieval
- Implements MCP protocol communication using the Agents framework
- Leverages Durable Objects for state management and consistency
- Uses Cloudflare Workers AI for generating embeddings
- 0 GitHub stars
Use Cases
- Personalizing chatbot interactions based on past conversations
- Providing context-aware assistance by retrieving relevant information from stored memories
- Remembering user preferences and behaviors for a more tailored experience