thedotmack/claude-mem: Persistent Context Across Sessions for Every Agent

What is Claude-Mem? Every time you start a new coding session with an AI agent, it starts from absolute zero. It re-reads the same files, asks the same questions, and falls back into mistakes you've already fixed together. Claude-Mem acts as a durable memory layer that sits beside your agent, silently taking notes on every read, edit, and command execution. How It Works and Core Features The system runs a local worker service powered by hooks. A secondary AI observer turns activity streams into concise observations and summarizes them when a session wraps up. When you open a new session, your agent receives a compact index of past work, fetching full details only when necessary. This saves tokens and keeps context sharp. It also provides advanced MCP search tools and a local web viewer for tracking your progress. Installation and Getting Started Getting started takes just a single command in your terminal. By running npx claude-mem install, the tool automatically discovers the supported AI agents on your machine, configures the necessary plugins, and starts the background worker. Detailed guides, custom modes, and configuration options are fully documented in the official repository.