Claude-Mem is a persistent memory compression system built specifically for Claude Code to preserve context across coding sessions. It automatically captures Claude’s tool usage, observations, and decisions, then compresses them into semantic memories that carry forward into future sessions. By enabling long-term continuity, Claude-Mem helps Claude “remember” project history, past fixes, and prior reasoning even after restarts or reconnects. Its progressive disclosure approach intelligently injects only the most relevant context, balancing usefulness with token efficiency. Claude-Mem runs automatically in the background with no manual workflow changes required. Designed for serious developers, it transforms Claude Code into a continuously learning, project-aware coding assistant.
Features
- Persistent memory that survives across Claude Code sessions
- Automatic capture and AI-driven compression of tool usage and observations
- Progressive disclosure system with layered, token-efficient context retrieval
- Natural-language memory search via the mem-search skill
- Web-based memory viewer UI with real-time streaming and citations
- Hybrid search using SQLite, FTS5, and Chroma vector embeddings
- Fine-grained privacy and context injection controls, including private tags
- Beta channel for experimental features like Endless Mode memory architecture