self-improvement

You log AI corrections and errors by writing entries into structured markdown files within a .learnings directory. This creates a persistent knowledge base that tracks failures and knowledge gaps to improve future interactions.

The best way to promote high-value learnings is to integrate them into project memory files like CLAUDE.md, AGENTS.md, or Copilot prompts. This ensures the captured knowledge actively guides future AI sessions.

Yes, you can use this skill to track knowledge gaps and feature requests. It captures these elements alongside user corrections and failures, organizing them in markdown logs for later extraction.

You apply captured learnings to OpenClaw or agent workflows by using the provided SKILL.md guidance to organize entries and promote them. This integrates your historical corrections directly into the active agent environment.

Yes, you need a .learnings log directory to start capturing learnings. The skill requires these structured markdown files, such as LEARNINGS.md, to exist so it can properly organize and promote your session data.