MemCP

소개

MemCP offers a robust solution to the perennial problem of context loss in Claude Code sessions. By implementing a Recursive Language Model (RLM) framework, it provides Claude with a persistent external memory, allowing it to store critical insights, decisions, and large documents on disk. This significantly reduces token consumption, enhances recall across sessions, and enables sophisticated knowledge navigation through a MAGMA 4-graph, ensuring that Claude never loses valuable context again. It features 21 MCP tools for memory and context management, sub-agent orchestration, and multi-tier search capabilities, all designed for seamless integration and improved LLM performance.

주요 기능

• 0 GitHub stars
• Persistent knowledge graph (MAGMA 4-graph) with semantic, temporal, causal, and entity edges
• Context-as-Variable (RLM) for token-efficient content storage and retrieval
• Automated context lifecycle management with auto-save hooks and multi-zone retention
• Five-tier search across memory insights and context chunks (keyword, BM25, fuzzy, semantic, hybrid)
• Orchestrates four Claude Code sub-agents for parallel processing and entity extraction

사용 사례

• Maintaining long-term context and knowledge across multiple Claude Code sessions
• Efficiently managing and navigating large documents without exceeding LLM context window limits
• Building complex LLM applications with modular sub-agents and persistent memory