memory-manager

Manage AI agent memory across four layers with file commands.

Updated Apr 26, 2026
One-click install
npx skills add https://github.com/2017java/super-agent-memory-manager --skill memory-manager-2017java
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: memory-manager
Source: https://github.com/2017java/super-agent-memory-manager/tree/main
Command: npx skills add https://github.com/2017java/super-agent-memory-manager --skill memory-manager-2017java

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires pypdf, pdfplumber, pdf2image, and includes scripts (resource) and references (resource) and assets (resource) components.

What problem does it solve?

This Skill provides a comprehensive solution for managing AI agent's memory, enabling persistent learning and preventing information loss across sessions.

Core Features & Use Cases

  • Layered Memory Architecture: Manages four layers of memory including Identity, User Profile, Candidate Pool, and Long-Term Memory.
  • Memory Management Commands: Offers a suite of commands to initialize, write, update, and manage memory files.
  • Candidate Pool: Collects observations and prevents mis-learning by requiring confirmation for persistent rules.
  • Hook Automation: Integrates with OpenClaw hooks for automated memory updates and error detection.
  • Multi-Agent Safety: Ensures safe collaboration across multiple agents by requiring approval for memory changes.
  • Use Case: Ideal for scenarios where AI agents need to maintain a knowledge base, such as customer support, technical assistance, or knowledge-intensive tasks.

Quick Start

Run 'memory-manager init' to set up the memory structure for your project. Use 'memory-manager write-memory' to add or update rules in the MEMORY.md file.

Frequently Asked Questions about memory-manager

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

FAQPage Schema
How do I implement persistent learning for an AI agent across sessions?

You implement persistent learning by initializing a 4-layer memory architecture that manages Identity, User Profile, Candidate Pool, and Long-Term Memory to prevent information loss across sessions.

How do I initialize and write rules to an AI agent's memory file?

You initialize the memory structure by running the 'init' command, then use the 'write-memory' command to add or update persistent rules directly in the MEMORY.md file.

What is a candidate pool in AI agent memory architecture and how does it prevent mis-learning?

A candidate pool in memory architecture collects observations and prevents mis-learning by requiring explicit confirmation before promoting temporary observations into persistent long-term rules.

Can I use hook automation for error detection and memory updates in multi-agent environments?

Yes, you can use OpenClaw hooks for automated memory updates and error detection, ensuring multi-agent safety by requiring explicit approval for any memory file changes.

Does this memory management approach support knowledge-intensive tasks like customer support?

Yes, this memory management approach is designed for knowledge-intensive tasks like customer support, maintaining a persistent knowledge base that retains user profiles and learned rules across interactions.

What are the limitations of using a candidate pool for AI agent memory management?

A limitation of using a candidate pool for memory management is that all new observations require manual confirmation before becoming persistent rules, which may bottleneck automation in fast-paced multi-agent workflows.