visual-verdict

Compare UI screenshots with reference images and output a deterministic JSON verdict.

1|Updated Apr 6, 2026
One-click install
npx skills add https://github.com/Hyeonjun0527/yeon --skill visual-verdict-hyeonjun0527
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: visual-verdict
Source: https://github.com/Hyeonjun0527/yeon/tree/main/.codex/skills/visual-verdict
Command: npx skills add https://github.com/Hyeonjun0527/yeon --skill visual-verdict-hyeonjun0527

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This skill compares a generated UI screenshot against reference images and returns a strict JSON verdict to drive the next edit iteration.

Core Features & Use Cases

  • Visual fidelity assessment across layouts, spacing, typography, and component styling
  • Deterministic pass/fail verdicts to guide design and development iterations
  • Flexible input handling with required reference_images[] and generated_screenshot, plus optional category_hint for targeted checks

Quick Start

Provide a generated_screenshot and one or more reference_images to obtain a JSON verdict

Frequently Asked Questions about visual-verdict

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I compare a generated UI screenshot against reference images for visual QA?▼

Visual QA comparison evaluates a generated UI screenshot against reference images to return a deterministic JSON verdict. The output enforces a fixed schema containing score, verdict, category_match, differences, suggestions, and reasoning to guide design iterations.

What is a deterministic JSON verdict for UI screen diffing?▼

A deterministic JSON verdict is a strict pass/fail assessment for UI screen diffs that returns a fixed schema. It includes score, verdict, category_match, differences, suggestions, and reasoning to provide objective guidance before applying design edits.

Can I use a category hint to target specific visual fidelity checks during image comparison?▼

Yes, you can supply an optional category_hint to target specific visual fidelity checks during image comparison. Along with required reference_images and a generated_screenshot, it helps focus the assessment on layouts, spacing, typography, or component styling.

Does visual regression testing for dashboards require a fixed JSON schema output?▼

Visual regression testing for dashboards uses a fixed JSON schema output to ensure deterministic pass/fail guidance. The schema enforces consistent fields like score, verdict, differences, and suggestions, which are necessary to drive automated design and development iterations.

What's the best way to automate pass/fail visual testing across app layouts?▼

The best way to automate pass/fail visual testing across app layouts is to generate a deterministic JSON verdict from a screenshot comparison. By supplying reference_images and a generated_screenshot, you receive strict scores and suggestions to automate the next edit iteration.

Why does my visual verdict output include reasoning and suggestions fields?▼

The visual verdict output includes reasoning and suggestions fields to provide actionable guidance for your next edit iteration. The fixed JSON schema is designed to return not just a pass/fail verdict and score, but also specific differences and recommendations to fix visual fidelity issues.