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...

This skill is largely coherent with its goal of capturing learnings, but it encourages storing extensive context — including file contents and command output — and provides hooks and scripts that, if enabled, will surface command outputs to the agent. Before installing or enabling hooks: 1) Review the scripts (activator.sh, error-detector.sh, extract-skill.sh and the hook handlers) to ensure they do only what you expect. 2) Avoid enabling user-global hooks unless you trust the source and have audited them; prefer project-level activation. 3) Be cautious about logging outputs that may include secrets (tokens, auth headers, private keys); modify templates/promotion policies to exclude or redact secrets and sensitive file paths. 4) If you must use extract-skill.sh, run it in a safe directory and review generated files before committing or promoting them. 5) Consider running in an isolated/dev environment first and add explicit filters to the error detector to avoid capturing sensitive lines. If you want a safer thumbs-up, provide confirmation that you (or a reviewer) have audited the scripts and will not promote raw file contents or secret-bearing outputs into shared workspace files.