Self Improving Agent

Captures learnings, errors, and corrections to enable continuous improvement. Use when: (1) A command or operation fails unexpectedly, (2) User corrects Clau...

Before enabling: 1) Review the three scripts (activator.sh, error-detector.sh, extract-skill.sh) and the hook handlers to confirm they do only what you expect — activator prints reminders; error-detector parses CLAUDE_TOOL_OUTPUT and prints a notice; extract-skill.sh writes files. 2) Pay special attention to the Error Entry template which suggests including "Full source of all included files" — avoid copying full source or any secret/config contents into .learnings/ or promoted files unless you have sanitized them. 3) If you enable PostToolUse hooks, use matcher filters or limit them to non-sensitive contexts so command outputs aren't broadly captured. 4) Test in an isolated workspace (no secrets, minimal repo) and use --dry-run where available before enabling globally. 5) Prefer manual invocation over global hooks if you cannot guarantee that logs will be sanitized. 6) Verify the skill origin (the README references multiple repos/forks) and only install from a trusted source. If you want a safer setup, disable automatic hooks and use the templates manually to log learnings.

Type: OpenClaw Skill Name: xaviermary56-self-improving-agent Version: 1.0.0 The skill bundle provides a structured framework for an AI agent to implement a 'self-improvement' loop by logging errors, user corrections, and new insights into markdown files. It includes utility scripts (e.g., `extract-skill.sh`, `error-detector.sh`) and OpenClaw hooks (`handler.js`) that inject reminders into the agent's context to capture knowledge. The logic is transparent, follows the stated purpose of persistent memory management, and contains no evidence of malicious intent, data exfiltration, or unauthorized execution.