A complete persistent memory system that eliminates context amnesia by giving AI agents long-term recall, identity files, and daily logs.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install memory-os
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install memory-os using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Memory OS (Persistent AI Agent Memory)?
Memory OS is a structural framework designed to solve the recurring problem of context amnesia in AI agents. By implementing a standardized set of markdown and JSON files within your workspace, this skill ensures your agent remembers your preferences, project history, and specific goals across multiple sessions. It transforms a standard LLM into a persistent collaborator that wakes up with full awareness of previous interactions.
This system integrates seamlessly with Openclaw Skills to provide a robust foundation for agentic workflows. It includes specialized templates for agent identity (SOUL.md), user profiles (USER.md), and automated heartbeat checks, ensuring that your AI assistant remains aligned with your working style and evolving project needs without requiring constant re-explanation.
Memory OS (Persistent AI Agent Memory) Use Cases
- Maintaining project continuity across disconnected AI chat sessions.
- Defining a permanent persona and set of expertise for your AI agent.
- Keeping a running log of daily progress and task status for complex development cycles.
- Storing user-specific preferences and style guides that the agent must always follow.
- Implementing proactive heartbeat checks to monitor project health or inbox status.
How Memory OS (Persistent AI Agent Memory) Works
- The system performs a mandatory security audit to ensure no external network calls or destructive commands are present.
- It detects the local development environment (such as Cursor, Windsurf, or OpenClaw) to apply platform-specific formatting adaptations.
- A dry run is presented to the user, listing all files to be created and confirming no existing data will be overwritten.
- Upon explicit confirmation, the skill initializes the directory structure and core configuration files including SOUL.md and USER.md.
- The agent creates an initial daily note and initializes the heartbeat state to track session-to-session continuity.
- The agent updates a local blueprint state file to manage the installation lifecycle and prepare for future Openclaw Skills updates.
Memory OS (Persistent AI Agent Memory) Setup
To install the Memory OS skill, provide the SKILL.md content to your agent and follow the interactive confirmation prompts.
# Ensure you are in your desired project workspace.
# No external package managers are required.
# Ask your agent: "Review the Memory OS blueprint and prepare for installation."
The installation will automatically configure the startup protocol and create the memory directory for persistent logs. Once installed, manually edit SOUL.md and USER.md to personalize your agent experience.
Memory OS (Persistent AI Agent Memory) Data Schema & Taxonomy
Memory OS organizes data using a flat-file structure for maximum portability and human readability across Openclaw Skills environments:
| File Path | Purpose | Creation Mode |
|---|---|---|
SOUL.md |
Defines agent identity, personality, and expertise. | Create (No Overwrite) |
USER.md |
Stores operator profile, goals, and working style. | Create (No Overwrite) |
MEMORY.md |
Persistent long-term knowledge and project history. | Create (No Overwrite) |
AGENTS.md |
Protocol for loading context at the start of a session. | Create (No Overwrite) |
HEARTBEAT.md |
Configuration for proactive agent checks and monitoring. | Create (No Overwrite) |
memory/*.md |
Daily session logs titled by date (YYYY-MM-DD). | Create (Daily) |
Memory OS (Persistent AI Agent Memory) Advanced Features
- Multi-platform adaptation logic for specialized IDEs and Openclaw Skills runners.
- Automated session startup protocol via AGENTS.md to ensure the agent reads memory files immediately upon activation.
- Proactive heartbeat system for monitoring external states like task lists or project milestones.
- Idempotent installation logic that prevents overwriting existing personalization files while allowing for state updates.
- Integrated memory maintenance workflow that distills daily logs into long-term knowledge assets automatically.
SKILL.md
Loading
Related Openclaw Skills
Inbox Triage
A professional email management skill for AI agents that scans inboxes, categorizes messages by urgency, and prepares draft replies using CLI-based email tools.
Content Calendar
A structured AI-driven system to plan, schedule, and track content pipelines across newsletters, blogs, and social media without spreadsheet chaos.
Writing Assistant
An AI writing partner that drafts and edits content while maintaining your unique brand voice and style preferences.
Research Assistant
A high-performance research protocol that enables AI agents to conduct multi-source web investigations, evaluate source quality, and maintain a structured knowledge library.
Project Tracker
An automated system for tracking multiple projects with status dashboards, milestone monitoring, and intelligent stalled-project detection.
Security Hardening
An automated security audit and workspace hardening suite designed to protect AI agents from credential leaks, PII exposure, and prompt injection attacks.
Featured*
