ThinkMem
Manages AI memory for Large Language Models, enabling them to better organize and utilize information through various storage types and communication protocols.
소개
ThinkMem is an innovative AI memory management system designed specifically for Large Language Models (LLMs). It functions as a Model Context Protocol (MCP) server, offering diverse memory types like unstructured RawMemory and structured ListMemory (arrays, queues, stacks) to enhance LLM capabilities. The system provides intelligent retrieval, automatic summary generation, and robust persistence options via JSON file storage. Supporting both standalone (stdio) and scalable HTTP modes, ThinkMem empowers LLMs to efficiently store, retrieve, and process information, thereby significantly improving their reasoning and recall abilities.
주요 기능
- Dual operating modes: MCP stdio and StreamableHTTP
- 1 GitHub stars
- Intelligent retrieval with text search and row-level operations
- Automatic summary generation and management
- Persistent JSON file storage with backup and recovery support
- Multiple memory types: RawMemory (unstructured text) and ListMemory (arrays/queues/stacks)
사용 사례
- Standalone memory management for individual AI assistants with explicit storage paths
- Centralized memory management for multiple AI assistants in a multi-machine setup via HTTP server
- Integration with AI assistants like Claude through MCP Host for enhanced memory capabilities
