AI Briefingclaude-mem: Claude Code Memory: Maintaining Context After Session End
thedotmack/claude-mem
About the project
Solves the problem of losing previous work context when a Claude Code session ends. It automatically captures tool usage records, converts them into semantic summaries, and injects this information into the next session. There is no need to manually leave notes or reconstruct context.
For token efficiency, it offers a 3-step search workflow. First, check the index with a search, then grasp the context via a timeline, and finally retrieve details only for necessary items. This filtering before fetching detailed data results in approximately 10x token savings.
It combines SQLite and Chroma vector databases to support hybrid search. Sensitive content can be excluded from storage using the private tag, and a real-time memory stream can be viewed via a web viewer. It also provides a skill to search memories in Claude Desktop.
thedotmack/claude-mem
Persistent Context Across Sessions for Every Agent – Captures everything your agent does during sessions, compresses it with AI, and injects relevant context back into future sessions. Works with Claude Code, OpenClaw, Codex, Gemini, Hermes, Copilot, OpenCode + More
JavaScript
This introduction was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report errors, attribution issues, or removal requests via Contact.