openclaw-persistent-memory

What This Skill Does

The openclaw-persistent-memory skill provides a robust, state-preserving layer for the OpenClaw AI agent. By leveraging a SQLite database enhanced with FTS5 (Full-Text Search), this skill enables the agent to remember context, user preferences, and specific project details across different sessions. Rather than relying on volatile context windows alone, this system captures key observations automatically, stores them in a structured local database, and intelligently retrieves relevant information using semantic and full-text indexing before responding to new prompts. This ensures the AI agent maintains continuity, reduces the need for redundant input, and provides a deeply personalized experience that grows more useful over time.

Installation

1. Install the global package via npm: npm install -g openclaw-persistent-memory.
2. Initialize the worker service by running openclaw-persistent-memory start.
3. Install the extension into your local directory: cp -r node_modules/openclaw-persistent-memory/extension ~/.openclaw/extensions/openclaw-mem.
4. Configure the integration by updating your ~/.openclaw/openclaw.json file. Ensure the slots memory key points to openclaw-mem and set autoCapture and autoRecall to true for the best results.
5. Complete the setup by restarting the OpenClaw gateway to register the new persistence layer.

Use Cases

• Project Continuity: Maintain complex project requirements, coding standards, or architectural decisions across multiple coding sessions without re-explaining context.
• Preference Management: Remember user-specific preferences such as preferred coding styles, communication tones, or formatting rules.
• Knowledge Retrieval: Use the agent as a personal knowledge base where it tracks discussions, meeting summaries, or research notes.
• Long-term Strategy: Track multi-day tasks or iterative improvements, allowing the agent to reference past successes or mistakes when generating future work.

Example Prompts

• "Based on my previous projects, can you summarize the naming convention I decided to use for the API endpoints?"
• "I remember we talked about using Tailwind CSS for the landing page last week, can you recall the specific color palette we settled on?"
• "Store this as a memory: My company's staging server is located at https://staging.example.com and requires the 'internal' VPN to access."

Tips & Limitations

To maximize the performance of your memory, be explicit when you want the AI to remember a specific fact; while auto-capture is effective, manual memory_store calls are more precise. Regularly monitor your memory usage by checking the /api/stats endpoint. Keep in mind that as the database grows, performant queries depend on the quality of your descriptors. Finally, avoid storing sensitive credentials or private keys in the memory store, as these are saved locally on your filesystem in a SQLite database.