memoria

What This Skill Does

Memoria is the foundational persistent memory layer for your OpenClaw agent. It transforms your AI from a stateless chatbot into a long-term collaborator that remembers your project history, professional background, technical preferences, and critical decision-making logic. By maintaining a structured Markdown file on your local machine, Memoria ensures that your agent remains context-aware across every session. It eliminates the "blank slate" problem, allowing the agent to reference previous lessons, ongoing development tasks, and specific project goals without requiring you to repeat yourself. Because it operates entirely locally, all your data remains private and strictly under your control, free from cloud storage or external APIs.

Installation

To install Memoria, run the following command in your terminal: clawhub install openclaw/skills/skills/contrario/agent-memoria

After installation, perform the one-time security setup to secure your data file:

1. Create the directory: mkdir -p ~/.memoria
2. Set restrictive permissions: chmod 700 ~/.memoria
3. Initialize the file: touch ~/.memoria/memory.md
4. Set file permissions: chmod 600 ~/.memoria/memory.md
5. Ignore the file in Git: echo ".memoria/" >> ~/.gitignore and echo "memoria.md" >> ~/.gitignore

Use Cases

• Project Continuity: Pick up exactly where you left off in complex software builds, with the agent remembering your architectural decisions and stack preferences.
• Learning Journal: Automatically track lessons learned during coding sessions or experiments to avoid repeating past mistakes.
• Preference Automation: Define your preferred tone, detail level, and code formatting styles once, and ensure the agent applies them consistently in every future response.
• Contextual Brainstorming: Refer back to previous ideas, blocked tasks, and recurring problems to maintain high-level project visibility.

Example Prompts

1. "Check my memory file and list the top three priorities for my current project marked as active."
2. "I'm starting a new feature implementation; review my 'Decisions Log' and 'My Stack' sections to ensure I don't contradict earlier design choices."
3. "Update the 'Lessons Learned' section with the fact that we fixed the memory leak by switching from async to sync logging."

Tips & Limitations

• Be Diligent: Memoria works best when you keep the file updated. The agent can suggest updates, but you should review the memory.md file periodically to prune obsolete data.
• Privacy: Since files are local, they will not sync across devices automatically unless you use a manual synchronization method. Always ensure your backup strategy remains compliant with your privacy requirements.
• Structure: Stick to the predefined Markdown headings (WHO I AM, WHAT I'M BUILDING, etc.) to ensure the agent can reliably parse and utilize your stored knowledge.