agent-memory-patterns

What This Skill Does

agent-memory-patterns is a sophisticated framework designed to provide OpenClaw agents with persistent, long-term memory capabilities. It moves beyond ephemeral session memory by implementing a hierarchical storage system that separates daily logging, pending tasks, and curated long-term knowledge. The system utilizes local file-based storage, allowing for human-readable and machine-searchable documentation. Through automated shell scripts, the agent can initialize daily logs, structure event metadata, and perform weekly curation of high-importance memories. By integrating this pattern, your agent transforms from a stateless assistant into an evolving entity capable of recalling past interactions, technical learnings, and specific project requirements across sessions.

Installation

To integrate this memory architecture into your OpenClaw agent, execute the following command in your terminal:

clawhub install openclaw/skills/skills/byron-mckeeby/agent-memory-patterns

Ensure that the agent process has read/write permissions for the ~/.openclaw/workspace directory, as the system relies on local disk I/O to maintain the memory/ subdirectory and the root MEMORY.md file.

Use Cases

• Project Management: Automatically tracking requirements and decisions made during daily stand-ups or technical deep-dives.
• Continuous Learning: Identifying recurring technical obstacles or code patterns learned during debugging sessions and formalizing them into a long-term knowledge base.
• Knowledge Management: Managing unstructured external content by passing data into pending-memories.md for later processing and filtering.
• Compliance & History: Maintaining a detailed, searchable audit trail of agent-user interactions for long-term project auditing.

Example Prompts

1. "Check my daily memory log for today and summarize the high-importance events concerning the database migration project."
2. "Perform the weekly memory curation task; please extract all insights tagged with 'high' importance from the past seven days and add them to my MEMORY.md."
3. "I just learned a new optimization trick for Python performance; please log this as a 'learned' event in today's daily memory file."

Tips & Limitations

• File Management: Since this skill performs heavy file operations, ensure your environment supports local persistent storage. Cloud-based containers that reset file systems on restart are not ideal unless they mount persistent volumes.
• Regex Limits: The current curation logic relies on grep and simple character markers. Ensure you strictly follow the marking format (🔴, 🟡, etc.) to ensure the curation scripts capture your most critical data accurately.
• Manual Cleanup: While the curation process automates much of the work, periodic review of MEMORY.md by a human is recommended to prune irrelevant data and maintain clarity.