What problem does it solve? Applying a skill improvement plan by hand is error-prone: recommendations span multiple SKILL.md files, reflections must be archived only after successful edits, and frontmatter validity must be preserved. This Skill automates that plan-application workflow so improvements land consistently across canonical source, bundle, and installed runtime tiers. ## Core Features & Use Cases - Plan-driven editing: Parses improvement plans produced by the skill-improvement-planner, groups recommendations per target skill, and applies them with apply_patch or the session's file-editing tool. - Scoped, validated edits: Edits only plan-named <harness>-* skills, supports --allow-skill allowlists and --dry-run, and validates YAML frontmatter, name, description, and referenced files after each change. - Reflection lifecycle management: Archives consumed reflections only after all citing recommendations succeed, and writes repo-local handoffs for the next pipeline stage. - Use Case: After a skill-improvement-planner run produces plan-v3.md, invoke this Skill to apply its recommendations to the targeted skills, archive the consumed reflections, and report the required bundle-regeneration follow-up. ## Quick Start Apply the latest improvement plan from the skill-improvement-planner to the targeted harness skills and archive the reflections it consumed.
