openclaw-persistent-memory

What This Skill Does

The openclaw-persistent-memory skill provides a robust, long-term memory layer for the OpenClaw AI agent. By utilizing a local SQLite database enhanced with FTS5 (Full Text Search), it enables the agent to store, retrieve, and reflect upon information gathered across multiple distinct sessions. Unlike standard conversational history which may be limited by context windows, this skill explicitly stores important user observations and insights, ensuring that critical data persists indefinitely. It features automated capture to reduce manual overhead and intelligent recall to inject relevant past context into the agent's current workflow, making the AI feel more personalized and context-aware.

Installation

Installation is a multi-step process. First, install the package globally via npm with npm install -g openclaw-persistent-memory. Next, initialize the background worker service by running openclaw-persistent-memory start. Integrate it into OpenClaw by copying the extension files to ~/.openclaw/extensions/openclaw-mem and running npm install within that directory. Finally, update your ~/.openclaw/openclaw.json configuration file under the plugins section to enable the openclaw-mem slot, specifying the worker URL and enabling auto-capture and auto-recall features before restarting your OpenClaw gateway.

Use Cases

This skill is ideal for power users who want their AI agent to remember project-specific requirements, user preferences, or complex technical details across weeks of development. It is particularly effective for managing ongoing coding projects where you want the agent to remember your specific coding style, preferred libraries, or previous roadblocks you have encountered in the same repository.

Example Prompts

1. "Remember that I prefer using Tailwind CSS for all my frontend projects instead of traditional CSS files."
2. "Search through your memories to find the notes I took last week about the API authentication schema for the project."
3. "Based on our previous conversations about the backend architecture, what did we decide regarding the database migration strategy?"

Tips & Limitations

To maximize the utility of this skill, ensure you explicitly tell the agent when information is important to store. While auto-capture is helpful, it may occasionally store noise. Periodically audit your observations using the provided REST endpoints to keep your memory bank clean. Note that since this relies on local SQLite storage, ensure your disk permissions are set correctly to avoid read/write errors. The system is designed for context retrieval, so keep entries concise to maintain low latency during semantic searches.