memory-system-v2

What This Skill Does

The memory-system-v2 provides a lightweight, performant, and persistent memory architecture for OpenClaw agents. By utilizing a file-based storage engine powered by bash and jq, it enables agents to store, categorize, and retrieve crucial information without the overhead of a database. The system excels at semantic storage, allowing the agent to recall learnings, decisions, insights, and events that occurred across different working sessions. Key features include sub-20ms search times, automated weekly consolidation, and an importance-scoring mechanism that helps prioritize relevant context during future interactions.

Installation

To install this skill, ensure that the 'jq' dependency is installed on your system (e.g., via 'brew install jq'). Once ready, you can add it to your environment by running:

clawhub install openclaw/skills/skills/kellyclaudeai/memory-system-v2

Ensure that the memory-cli.sh script is executable and accessible within your workspace path.

Use Cases

• Continuous Project Tracking: Maintain a record of architectural decisions and trade-offs made during long-term coding projects.
• Learning Management: Keep a structured log of new concepts, coding patterns, or tools learned during research, ensuring they aren't forgotten.
• Insight Synthesis: Capture 'aha' moments and breakthrough ideas during brainstorming sessions to be consolidated and reviewed later.
• Automated Reporting: Use the consolidation feature to summarize a week's worth of progress and learnings into a concise document.

Example Prompts

1. "Capture a new learning: I found that using React Query significantly reduces our boilerplate code for data fetching in our frontend apps. Set importance to 8."
2. "Search my memory for any insights I recorded regarding 'user interface' or 'accessibility' from last month."
3. "Consolidate my memories for this week and provide a summary of the most important decisions I made."

Tips & Limitations

• Be Specific with Tags: The search performance relies on effective tagging. Use consistent, descriptive tags to make retrieval faster and more accurate.
• Importance Matters: Assigning correct importance scores (1-10) is vital; higher scores will bubble up in priority during searches.
• Cleanup: While the system supports auto-consolidation, periodically checking your memory store keeps the file-based system efficient.
• Limitations: Since it is file-based, it is best suited for individual agent memory rather than massive enterprise-scale datasets; large volumes of data may eventually require structured database migration.