AI Agent Hub
Back to skills
Skill Quality Reviewer icon

Skill Quality Reviewer

AI Agent Updated 2026.08.30

Paste the following prompt into your AI chat to install this skill:

Please follow https://skillhub.cn/install/skillhub.md to install @user_f49a700e/skill-opt-reviewer.

About this skill

What It Solves

Many SKILL.md files only expose problems after release: a too-broad or too-narrow description can misfire, missing examples make model behavior unstable, and mixed resources make review noisy. skill-reviewer moves this check earlier by reading the target Skill's SKILL.md, directory structure, and related assets, then producing a review report against a fixed checklist.

How It Works

  • Collects files: reads SKILL.md, lists scripts/, references/, assets/, and reads relevant resources.
  • Loads standards: uses 100 base points and 30 bonus points from references/checklist.md.
  • Scores item by item: highlights only non-perfect checks with targeted explanations.
  • Outputs a report: dimension table, bonus score, anti-patterns, improvement items, and summary.
  • Optionally generates test cases: positive, negative, and boundary scenarios, written to evaluation/test-cases.md.

Boundaries

It is best for pre-release review and code-style improvement, not for replacing model capability testing or business acceptance. Scoring depends on checklist.md and examples.md; for highly specialized domains, humans should still judge whether deductions are too strict or too loose.

Use Cases

  • Before release, review the description and steps and produce an improvement report.
  • When a team shares a skill folder, check structure, assets, and checklist scores.
  • After editing an existing skill, compare base and bonus scores to find deductions.
  • Before launch, generate positive, negative, and boundary cases and write them to a test file.

Best For

  • Agent engineer maintaining shared skills who wants to catch unstable triggers and structure issues before release.
  • Full-stack engineer managing a team skill repo who needs consistent checklist scores and improvement items.
  • AI product engineer packaging agent assets who needs review reports and test cases saved into the repo.
  • Engineering owner taking over legacy skills who wants quick base, bonus, and anti-pattern findings.