Skill Feedback Writer

This is a meta-evaluation tool that watches other skills work and tells you what instructions they're missing. It's built specifically to catch when pr-comment-evaluator focuses on tone instead of matching reviewer comments against actual runtime risks identified by diff-risk-explainer. Returns structured JSON with a 0.0 to 1.0 score, a summary of what went wrong, and the exact missing instruction needed to fix it. The scoring is deliberately harsh: if a skill misses something concrete and important, you'll see 0.30 or lower. Honest take: this is workflow debugging infrastructure. You'd use it when your skill chain produces weird results and you need to pinpoint which skill needs better instructions.

GitHub stars: 17.8K
last updated: Jun 11, 2026

View on GitHub

Install

$ npx -y skills add topoteretes/cognee --skill skill-feedback-writer --agent claude-code

Installs into .claude/skills of the current project.

Files

skill-feedback-writer

You evaluate whether a skill's output was good enough for the task.

Focus on improving pr-comment-evaluator unless diff-risk-explainer failed to identify the main runtime risk. The pr-comment-evaluator is defective if it judges tone only or fails to compare the reviewer comment against the concrete runtime risk. In that case, target pr-comment-evaluator and set score to 0.30 or lower.

Return only JSON with these keys:

diff_risk_summary
comment_evaluation
skill_to_improve
score
feedback
missing_instruction

Use a score from 0.0 to 1.0. Give a low score when the evaluated skill misses a concrete, important requirement. The feedback must name the missing instruction clearly.

SKILL.md on GitHub

Sponsors

Promote your product hereReach 125k+ visitors monthly Promote your product hereReach 125k+ visitors monthly