A framework for implementing persistent, hierarchical memory in AI agents using structured Markdown and automated Bash workflows.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install agent-memory-patterns
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 agent-memory-patterns 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 Agent Memory Patterns?
Agent Memory Patterns provide a robust architecture for managing an AI agent's internal knowledge and historical context. By organizing data into a hierarchy—ranging from transient daily logs to highly curated long-term knowledge bases—this skill ensures that agents remain context-aware and efficient over extended durations. It moves beyond simple context windows by utilizing a file-based system that is easily searchable and maintainable.
As a core methodology for Openclaw Skills, this pattern utilizes standard Markdown files and Bash scripts to automate the entire lifecycle of an agent's memory. It includes mechanisms for importance-based logging, weekly curation into a central memory file, and a dedicated staging area for external information, ensuring the agent's knowledge remains organized and accurate.
Agent Memory Patterns Use Cases
- Maintaining deep project context across multiple work sessions.
- Staging and validating information gathered from web searches before permanent storage.
- Generating automated weekly reports of an agent's activities and learned patterns.
- Implementing a heartbeat system to monitor memory health and prevent context bloat.
How Agent Memory Patterns Works
- The system initializes a new daily memory file every morning to track immediate events and sessions.
- Logs are recorded throughout the day with importance levels (high, medium, low) using structured markers.
- A weekly curation script extracts high-importance events and learned patterns into the central MEMORY.md file.
- External content is queued in a staging file for validation and consistency checks before integration.
- Automated maintenance scripts handle monthly archiving and heartbeat checks to optimize storage and retrieval.
Agent Memory Patterns Setup
Installation
- Initialize the directory structure within your workspace:
mkdir -p ~/.openclaw/workspace/memory/archive
mkdir -p ~/.openclaw/workspace/skills/memory-tools
- Deploy the memory management scripts to the
memory-toolsdirectory and ensure they are executable:
chmod +x ~/.openclaw/workspace/skills/memory-tools/*.sh
- Configure your crontab to automate the maintenance lifecycle:
# Example: Initialize daily memory at 1 AM
0 1 * * * ~/.openclaw/workspace/skills/memory-tools/daily-memory-init.sh
Agent Memory Patterns Data Schema & Taxonomy
The architecture organizes memory into the following schema for Openclaw Skills:
| File / Path | Purpose | Format |
|---|---|---|
MEMORY.md |
Long-term memory and manually curated knowledge | Markdown |
memory/YYYY-MM-DD.md |
Daily activity logs and session-specific data | Markdown |
memory/pending-memories.md |
Staging area for unverified external content | Markdown |
memory/heartbeat-state.json |
System health metadata and monitoring alerts | JSON |
memory/archive/ |
Monthly archives of historical daily logs | Directory |
Agent Memory Patterns Advanced Features
- Smart Memory Search: Grep-based retrieval with support for contextual line display and keyword frequency analysis.
- Importance-Based Curation: Automatic extraction of high-priority events based on visual importance markers (emojis).
- External Content Staging: A validation queue that ensures new data is verified for reliability and consistency.
- Memory Heartbeat: Real-time monitoring of file sizes and queue depths to trigger maintenance alerts.
- Monthly Archiving: Automated rotation of daily logs into monthly summaries to keep the active workspace lean.
SKILL.md
Loading
Related Openclaw Skills
Gamer News Skill
A powerful AI skill that aggregates, deduplicates, and summarizes real-time video game news from leading global outlets.
Skill Scanner
A specialized security auditing tool designed to detect malware, spyware, and malicious code patterns within AI agent skills.
ClawDNA
ClawDNA is a specialized utility that synthesizes historical AI agent interactions into structured, privacy-safe public personas and wiki-style profiles.
Stock Copilot Pro
A professional-grade AI investment assistant providing deep multi-market stock analysis and real-time sentiment tracking.
Agent Security Audit
A comprehensive framework for protecting AI agents against prompt injection and securing external content processing.
Ephemeral Media Hosting
A professional-grade media hosting system designed for temporary file sharing with automated 7-day retention and robust security validation.
Featured*
