OpenParable
Skip to content
← Back to skills

Refactor UI

Design
refactor-ui

Refactor UI is an Agent Skill for design published in the gnurio/refactoring-ui-plugin repository on GitHub: a SKILL.md instruction file an AI agent can load to do this task.

Analyze a design against all 10 Refactoring UI skills and generate a comprehensive assessment with specific fixes

View on GitHub ↗
QUICK START

How to use this skill

Bring this guide into your coding agent with a prompt tailored to the tool you use.

  1. Open your project in Codex.
  2. Copy and paste the prompt below.
  3. Review files and risks before approving any action.
Prompt to paste
Inspect the Agent Skill at https://github.com/gnurio/refactoring-ui-plugin/blob/HEAD/skills/meta-refactor-ui/SKILL.md and all supporting files. Report provenance, permissions, scripts, dependencies, and security risks. Propose an exact installation into .agents/skills/refactor-ui, but wait for my approval before writing files or running scripts.

Copying this prompt does not install or run the skill. Review third-party files before use.

Meta-Skill: /refactor-ui

Purpose

Run a complete UI design audit against all 10 Refactoring UI principles, generating a prioritized list of specific improvements.

Workflow

Phase 1: Load Skill Registry

Read skills.json
Validate all 10 skills available

Phase 2: Sequential Skill Application

Execute each skill in optimal order:

1. establish-visual-hierarchy
   ↓
2. apply-typography-scale
   ↓
3. build-color-palette
   ↓
4. apply-consistent-spacing
   ↓
5. design-button-hierarchy
   ↓
6. eliminate-visual-clutter
   ↓
7. design-empty-states
   ↓
8. use-shadows-appropriately
   ↓
9. manage-color-contrast
   ↓
10. group-related-elements

Phase 3: Consolidate Findings

Aggregate results from all skills:

  • Collect all FAIL assessments
  • Group by severity (Critical | High | Medium | Low)
  • Remove duplicates (same issue caught by multiple skills)
  • Prioritize by impact

Phase 4: Generate Report

Output structured assessment:

{
  "overall_score": "PASS | NEEDS_WORK | FAIL",
  "summary": "3 critical, 5 high, 2 medium priority fixes",
  "violations": [
    {
      "skill": "visual-hierarchy",
      "severity": "critical",
      "issue": "Primary CTA buried",
      "location": "Hero section",
      "fix": "Increase button to solid brand color, reduce surrounding contrast"
    }
  ],
  "priority_fixes": [
    "1. Make primary CTA prominent with brand color",
    "2. Consolidate font sizes from 12 to 6",
    "3. Add 8-10 grey shades to palette"
  ],
  "skill_breakdown": {
    "visual-hierarchy": { "status": "FAIL", "issues": 2 },
    "typography-scale": { "status": "PASS", "issues": 0 },
    "color-palette": { "status": "FAIL", "issues": 1 },
    ...
  }
}

Execution Modes

Mode: Quick Scan

  • Run all 10 skills
  • Report only FAILs
  • Time: ~30 seconds

Mode: Deep Analysis

  • Run all 10 skills
  • Report PASS with rationale
  • Cross-reference between skills
  • Suggest composition improvements
  • Time: ~2 minutes

Mode: Fix Mode

  • Run all 10 skills
  • Generate specific fix instructions
  • Provide before/after code examples
  • Time: ~5 minutes

Usage Examples

Example 1: Marketing Page

Input: Landing page description with hero, features, CTA

/refactor-ui
→ Overall: NEEDS_WORK
→ Critical: Visual hierarchy (CTA buried)
→ High: Typography (8 sizes, need 5)
→ Medium: Spacing (ambiguous grouping)
→ Priority fixes: [3 items]

Example 2: Dashboard

Input: Analytics dashboard screenshot

/refactor-ui --deep
→ Overall: PASS with suggestions
→ Visual hierarchy: PASS (clear primary)
→ Color palette: FAIL (only 4 greys)
→ Shadows: FAIL (decorative on cards)
→ Suggestions: [5 items]

Integration with Validation

This meta-skill can validate against NotebookLM:

For each skill assessment:
  Query: "According to Refactoring UI, is this correct?"
  Validate assessment matches expert
Report agreement rate

References

  • ../skills.json - Skill registry
  • ../examples/ - Before/after examples for all skills