crab-mem

Install persistent memory for OpenClaw agents with claude-mem worker.

17|7|Updated Feb 1, 2026
One-click install
npx skills add https://github.com/thedotmack/crab-mem --skill crab-mem
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: crab-mem
Source: https://github.com/thedotmack/crab-mem/tree/main/skills/crab-mem
Command: npx skills add https://github.com/thedotmack/crab-mem --skill crab-mem

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) components.

What problem does it solve?

This Skill installs persistent memory for OpenClaw agents, enabling long-term memory across sessions and improving agent continuity.

Core Features & Use Cases

  • Installs claude-mem worker service to run continuously
  • Configures the OpenClaw memory plugin for seamless integration
  • Captures observations on every tool call and injects context into new sessions
  • Provides verification and basic troubleshooting steps for reliability

Quick Start

Use the quick install to bootstrap memory for your OpenClaw agent: ./scripts/install.sh Or run directly: curl -fsSL https://crab-mem.sh | bash

Frequently Asked Questions about crab-mem

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I add persistent memory to OpenClaw agents so they remember context across sessions?

Persistent agent memory works by running a local worker service on port 37777 that captures observations on every tool call, stores them, and automatically injects that context into subsequent agent sessions.

What do I need to install persistent memory for an OpenClaw agent?

You need Node.js 18+, bun, git, curl, internet access to fetch claude-mem and plugin code, a working OpenClaw installation, and local port 37777 available for the worker service.

How do I quickly set up context injection for OpenClaw agents?

Run the quick install script via curl or execute ./scripts/install.sh to bootstrap the claude-mem worker, configure the OpenClaw memory plugin, and enable automatic observation capture across sessions.

Does automatic observation capture work without keeping a background service running?

No, automatic observation capture requires the claude-mem worker service to run continuously on local port 37777 to intercept tool calls and persist context for injection into new sessions.

What are the limitations of using a worker service for agent memory persistence?

Limitations include requiring local port 37777 availability, continuous execution of the claude-mem worker, and internet access during installation to fetch the necessary plugin code and dependencies.