Memory

概要

Memory is an MCP server designed to give AI assistants a true long-term memory, ensuring continuity across conversations, platforms, and sessions. Built to overcome the limitations of session-bound AI interactions, it leverages PostgreSQL with pgvector for efficient semantic search, allowing AI to recall contextually relevant information. The system supports flexible embedding models (Ollama, OpenAI, and compatible APIs), encrypts sensitive memories with AES-256-GCM, and offers robust authentication via OAuth and bearer tokens, all while isolating different memory sets using namespaces.

主な機能

• Secure encryption of sensitive memories (AES-256-GCM)
• 0 GitHub stars
• Flexible integration with embedding models (Ollama, OpenAI, OpenAI-compatible APIs)
• Namespaces for memory isolation and organization
• Persistent memory for AI conversations using PostgreSQL and pgvector
• OAuth and Bearer Token authentication for secure access

ユースケース

• Maintaining long-term conversational context for AI across multiple platforms and sessions
• Enabling AI assistants to recall and apply past interactions and information
• Storing and retrieving relevant conversational fragments based on semantic search queries