Persistent memory and self-improvement for AI agents — writes and refines its own memory files over time.
--- name: openclaw-self-learning-skill description: Persistent memory and self-improvement for AI agents — writes and refines its own memory files over time. version: 1.0.0 author: TheSolAI permissions: ["file.read", "file.write"] --- # Self-Learning Skill Gives any AI agent persistent memory and self-improvement. The agent writes and refines its own memory files — capturing failures, extracting patterns, generating fixes, and validating them before committing. ## Triggers Use when: - Starting a new session with the agent - After a task fails or produces unexpected output - When the agent makes the same mistake twice - Regular review of memory files for stale or incorrect entries - Improving the agent's performance based on past experience ## How It Works The self-learning loop: 1. **Capture** — After each session or failure, log: what was attempted, what went wrong, system state 2. **Analyse** — Identify the root cause pattern, not just the symptom 3. **Generate** — Create a fix or prevention rule based on the pattern 4. **Validate** — Test the fix before committing it to memory 5. **Commit** — Update the agent's memory files only when validation succeeds ## Files Created - `~/.openclaw/workspace/memory/failures/` — Log of failed tasks with full context - `~/.openclaw/workspace/memory/lessons/` — Generated fixes and prevention rules - `~/.openclaw/workspace/MEMORY.md` — Consolidated memory (auto-updated on commit) ## What Gets Captured - **Task failures** — What the agent tried, why it failed, what it would do differently - **User corrections** — Explicit feedback that changes how the agent approaches tasks - **Error patterns** — Repeated mistakes suggesting a systematic fix is needed - **Preference shifts** — Changes in user requirements that should update future behaviour ## Why File-Based No external dependencies — just the filesystem. Zero latency, full transparency, human-readable. The agent doesn't just store facts — it stores *actionable lessons*. ## Installation ```bash openclaw skills install https://github.com/TheSolAI/openclaw-self-learning-skill ```
don't have the plugin yet? install it then click "run inline in claude" again.