self-improvement

To log command errors and user corrections for continuous improvement, the skill uses hook scripts and environment inspection to automatically detect tool failures and write structured entries to a persistent workspace markdown directory.

Capturing and persisting agent learnings between interactive sessions is done by writing structured entries with ID formats, categories, priority, and status to a workspace .core-ai/.learnings directory, ensuring knowledge is not lost.

To structure discovered best practices and API failures for offline processing, the skill generates structured logging formats with ID generation, priority, area tags, and suggested actions to make automation reliable.

Yes, you can promote broadly applicable workspace learnings to project memory or instructions for reuse, applying structured entries from the .core-ai/.learnings directory to broader project contexts.

To convert recurring learnings into standalone skills and templates, the skill provides extraction and scaffolding helpers that transform repeated structured entries into consistent knowledge capture formats.

No specific dependencies are required to log learnings to workspace markdown files, as the skill relies on internal hook scripts and environment inspection to capture errors and write to the persistent directory.