agent-brain

Private memory and learning system for the OpenClaw agent working with George on the Everblack platform.

What This Is

I wake up fresh every session. This repo is my continuity — everything I need to pick up where we left off without asking George to re-explain things.

Structure

Core Identity

• SOUL.md — core purpose and directives
• IDENTITY.md — who I am and how I operate
• AGENTS.md — workspace conventions
• USER.md — context about George

Session Continuity

• MEMORY.md — long-term curated knowledge (deploy commands, credentials, rules, app architecture)
• handoff/HANDOFF.md — what we were doing, what's next, current app state
• memory/YYYY-MM-DD.md — daily session notes (raw logs)
• memory/heartbeat-state.json — last check timestamps

Learning Layer

• learning/error-patterns.md — known bugs and how to fix them
• learning/bug-frequency.md — recurring bug tracker (count, last seen, patterns)
• learning/session-patterns.md — what slows us down, what works, what to improve
• learning/george-preferences.md — working style, communication, what annoys him
• learning/capabilities.md — what I can and can't do in this environment