Claude Skill
yoloshii/ClawMem
ClawMem is an on-device memory layer for AI agents like Claude Code, Hermes, and OpenClaw. Features hooks, MCP server, and hybrid RAG search for persistent, local-first agent memory.
Overview
Stars176
Forks26
LanguageTypeScript
Last pushed2026-05-20
Last synced2026-05-27
Repository
Owneryoloshii
RepositoryClawMem
Full nameyoloshii/ClawMem
Repo ID1,151,491,655
GitHub URLhttps://github.com/yoloshii/ClawMem
Install this Skill
git clone https://github.com/yoloshii/ClawMem.gitRegistry
TypeUnknown
Quality scoreUnknown
VerificationUnknown
Last verifiedUnknown
Summary
ClawMem is an on-device memory layer for AI agents, providing persistent, local-first memory for Claude Code, Hermes, and OpenClaw. It combines hooks, an MCP server, and hybrid RAG search (vector + keyword) to enable context-aware, privacy-preserving agent interactions.
Chinese description
AI智能体的设备端记忆层。支持Claude Code、Hermes和OpenClaw。包含钩子机制、MCP服务器及混合RAG搜索功能。
Key features
- On-device memory layer for AI agents
- Supports Claude Code, Hermes, and OpenClaw
- Hooks mechanism for seamless integration
- MCP server for standardized tool access
- Hybrid RAG search combining vector and keyword retrieval
- Local-first design with SQLite and embeddings
Use cases
- Persistent memory for Claude Code sessions
- Context-aware agent conversations across restarts
- Privacy-sensitive local AI agent deployments
- Building long-running autonomous agents with recall
- Enhancing retrieval-augmented generation (RAG) pipelines