Claude Skill
NeoLi00/memX
memX is a self-learning, self-maintaining memory plugin for AI agents with native support for Claude Code, Codex, and OpenClaw. It uses embeddings and graph memory for long-term context.
Overview
Repository
Install this Skill
npx -y -p github:NeoLi00/memX memx quickstart claude-code \Registry
npx -y -p github:NeoLi00/memX memx quickstart claude-code \npx -y -p github:NeoLi00/memX memx quickstart codex \npx -y -p github:NeoLi00/memX memx quickstart openclaw \npx -y -p github:NeoLi00/memX memx quickstart mcp \npx -y -p github:NeoLi00/memX memx service status
Summary
memX is a self-learning, self-maintaining memory plugin for AI agents, offering native support for Claude Code, Codex, and OpenClaw. It enables agents to autonomously build, update, and retrieve long-term memory using embeddings and graph-based structures.
memX：面向AI代理的自我学习、自我维护记忆插件；原生支持Claude Code、Codex和OpenClaw
Key features
- Self-learning memory that evolves with agent interactions
- Self-maintaining mechanism for automatic memory cleanup and optimization
- Native integration with Claude Code, Codex, and OpenClaw
- Embeddings-based memory retrieval for relevant context
- Graph memory structure for relational knowledge representation
- Long-term memory persistence across sessions
Use cases
- Enhancing Claude Code with persistent project context
- Building AI coding assistants that remember user preferences
- Creating autonomous agents with evolving knowledge bases
- Supporting multi-session conversations in OpenClaw
- Improving Codex agent performance with long-term memory
README excerpt
<p align="center"> <img src="./assets/memx-cover-en.svg" alt="memX - self-learning, self-maintaining memory for AI agents" width="920"> </p> <p align="center"> <a href="./README.md">English</a> · <a href="./README-ch.md">中文</a> · <a href="./ARCHITECTURE.md">Architecture</a> </p> --- memX turns completed work into structured, searchable, self-maintained memory, then injects only the evidence an agent needs for the current query. It connects natively to Codex, Claude Code, and OpenClaw, and reaches any MCP-compatible client through the same local memory layer. ## Benchmarks <table align="center"> <thead> <tr> <th>Suite</th> <th>Scope</th> <th>R@3 success rate</th> </tr> </thead> <tbody> <tr> <td><strong>LongMemEval-S</strong></td> <td>Long-context memory retrieval</td> <td><strong>94.2%</strong></td> </tr> <tr> <td><strong>Real engineering cases</strong></td> <td>30 cases, each with 20+ turns</td> <td><strong>100%</strong></td> </tr> </tbody> </table> ## Architecture <p align="center"> <img src="./assets/memx-overview.svg" alt="memX coarse architecture" width="920"> </p> ## Agent support <table align="center"> <tr> <td align="center" width="56"><img src="./assets/agent-logos/codex.png" alt="Codex logo" width="34"></td> <td><strong>Codex</strong></td> <td><sub>native hooks, MCP hidden by default</sub></td> </tr> <tr> <td align="center" width="56"><img src="./assets/agent-logos/claude-code.png" alt="Claude Code logo" width="34"></td> <td><strong>Claude Code</strong></td> <td><sub>native hooks, MCP hidden by default</sub></td> </tr> <tr> <td align="center" width="56"><img src="./assets/agent-logos/openclaw.png" alt="OpenClaw logo" width