MCP Catalogs
HomeClawMem screenshot

ClawMem

by yoloshii·168·Score 50

On-device memory layer for AI agents with MCP server integration and hybrid RAG search.

ai-llmdeveloper-toolsknowledge-graph
26
Forks
1
Open issues
this month
Last commit
2d ago
Indexed

Overview

ClawMem is a comprehensive on-device memory system for AI agents including Claude Code, OpenClaw, and Hermes. It combines retrieval-augmented search with hooks, an MCP server, and hybrid RAG capabilities in a single local system. The architecture employs advanced techniques like QMD-derived multi-signal retrieval, SAME-inspired composite scoring, and MAGMA-style intent classification with multi-graph traversal. All components work together through a local SQLite vault, ensuring privacy and offline operation without requiring API keys or cloud dependencies.

Try asking AI

After installing, here are 6 things you can ask your AI assistant:

you:Enhancing AI coding agents with persistent memory of project decisions and context
you:Creating a unified memory layer across different AI agent platforms (Claude Code, OpenClaw, Hermes)
you:Implementing retrieval-augmented generation for personal notes and documentation
you:What platforms does ClawMem support?
you:Does ClawMem require API keys or cloud services?
you:How does ClawMem integrate with MCP?

When to choose this

Choose ClawMem if you need a local, persistent memory layer for AI coding agents that maintains context across sessions and works offline without cloud dependencies.

When NOT to choose this

Don't choose ClawMem if you prefer a cloud-based memory system or need integration with non-supported AI agents that don't have MCP plugins or hooks.

Tools this server exposes

3 tools extracted from the README
  • memory_retrieve

    Retrieve relevant memories from the vault based on query intent

  • timeline

    Navigate temporal neighborhoods around a document

  • focus_set

    Set a session-scoped focus topic to boost relevance

Note: Tool names inferred from descriptions in the README text, as no explicit 'Tools' section was provided. The memory_retrieve tool appears to be the core retrieval mechanism, while timeline and focus_set seem to be supporting tools for context

Comparable tools

mem-gptsemantic-workspacelangchain-mcpembedchain

Installation

Install via npm (recommended)

npm install -g clawmem

Install via Bun

bun add -g clawmem

Setup for Claude Code (MCP integration)

clawmem bootstrap
clawmem setup hooks
clawmem setup mcp

Verify installation

clawmem doctor

FAQ

What platforms does ClawMem support?
ClawMem fully supports Linux and macOS. Windows is supported via WSL2, but not recommended for native installation.
Does ClawMem require API keys or cloud services?
No, ClawMem runs fully local with no API keys and no cloud dependencies. All memory is stored in a local SQLite vault.
How does ClawMem integrate with MCP?
ClawMem exposes its functionality through an MCP server that works with any MCP-compatible client. After setup, tools are automatically available to your MCP client.

On Hacker News

Recent discussion from the developer community.

Compare ClawMem with

GitHub →

Last updated · Auto-generated from public README + GitHub signals.