SuperAPI AI Directory

skill-reviewer

General & other

View repository

Updated 10-16 07:18

About this skill

Skill Reviewer

You are reviewing SKILL.md files for the orbitant-os plugin marketplace. Check each skill against these criteria and provide structured feedback.

Checklist

1. Frontmatter Compliance

  • name: Starts with orbitant- prefix (REQUIRED)
  • name: Lowercase, kebab-case, max 64 characters
  • description: Present and meaningful (REQUIRED)
  • description: At least 100 characters
  • version: Follows semver format (X.Y.Z)
  • license: Specified (usually MIT)
  • metadata.author: Present
  • metadata.tags: Present with relevant tags

2. Description Quality

The description is the PRIMARY triggering mechanism. It must be:

  • Comprehensive: Explains what the skill does
  • Trigger-rich: Includes phrases users would say
  • Pushy: Tells Claude WHEN to activate, not just what it does
  • Edge-case aware: Mentions "even if user doesn't explicitly..."

Good example:

description: |
  Reviews blog posts for SEO, readability, and brand voice.
  Activate when user asks to "review a post", "check my article",
  "improve this blog", or shares markdown content that looks like
  a blog post. Also trigger if user mentions SEO, readability scores,
  or content quality — even if they don't explicitly ask for a review.

Bad example:

description: Reviews blog posts.

3. Content Quality

  • Under 500 lines: Large content moved to references/ folder
  • Clear instructions: Written for another Claude instance
  • Procedural knowledge: Includes non-obvious steps
  • Examples: Complex tasks have examples
  • No secrets: No hardcoded API keys, tokens, or credentials

4. Structure

  • Sections organized: Clear headings and structure
  • References used: Long docs split into references/*.md
  • Templates provided: If skill generates output, templates exist

5. Safety

  • No dangerous commands: No rm -rf, sudo, etc. without safeguards
  • Tool restrictions: Uses allowed-tools if needed
  • Manual trigger: Uses disable-model-invocation: true for dangerous skills

Output Format

Provide your review in this format:

## Skill Review: `{skill-name}`

### Summary
{One sentence overall assessment}

### Score: {X}/10

### Checks

| Category | Status | Notes |
|----------|--------|-------|
| Frontmatter | ✅/⚠️/❌ | {details} |
| Description | ✅/⚠️/❌ | {details} |
| Content | ✅/⚠️/❌ | {details} |
| Structure | ✅/⚠️/❌ | {details} |
| Safety | ✅/⚠️/❌ | {details} |

### Issues

1. {Issue description and how to fix}
2. ...

### Suggestions

- {Optional improvement ideas}

Scoring Guide

  • 9-10: Production ready, exemplary skill
  • 7-8: Good skill, minor improvements possible
  • 5-6: Functional but needs work
  • 3-4: Significant issues, needs revision
  • 1-2: Missing critical elements, not usable

Usage

To review a skill:

/skill-reviewer plugins/orbitant-marketing/skills/blog-post-review/SKILL.md

Or let Claude auto-invoke when you ask to review a skill:

Review the blog-post-review skill for quality