skill-improve

Run a test-fix-retest workflow to reduce static analysis failures and warnings by measuring baseline counts, diagnosing exact gaps tied to specific check IDs, and applying targeted edits to frontmatter and body instructions until warnings are resolved.

A test-fix-retest loop for skill quality remediation runs static and category rubric checks, proposes minimal before/after edits with confirmation, and keeps or reverts changes based on whether the combined validation score improves.

Improve code quality and reduce warnings in skill frontmatter by running baseline static checks, mapping failures to concrete frontmatter gaps, proposing minimal edits, and validating the updated file using static and category tests.

Yes, you can revert skill edits if the retest score does not improve, because the workflow automatically reverts changes when the combined score from the static and category rubric checks fails to show measurable improvement.

Yes, skill improvement automation requires manual confirmation before writing changes, because the workflow shows before/after proposals for targeted edits and asks for permission before modifying the skill file.

The best way to automate quality gate checks for skills in .claude/skills/ is running a measurable test-fix-retest workflow that executes static and optional category rubric checks, diagnoses failures, and validates improvements until the skill passes documented gates.