visual-verdict

Compare UI screenshots to reference images and output a JSON pass/fail verdict.

1|Updated Apr 1, 2026
One-click install
npx skills add https://github.com/OliverOuyang/shuhe-work-skills --skill visual-verdict-oliverouyang
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: visual-verdict
Source: https://github.com/OliverOuyang/shuhe-work-skills/tree/main/skills/visual-verdict
Command: npx skills add https://github.com/OliverOuyang/shuhe-work-skills --skill visual-verdict-oliverouyang

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill provides a deterministic, machine-readable verdict for comparing generated UI screenshots against one or more reference images, removing guesswork from visual review cycles and enabling automated edit iterations.

Core Features & Use Cases

  • Strict JSON Verdict: Returns a fixed JSON schema with score, verdict, category_match, differences, suggestions, and concise reasoning for deterministic decision-making.
  • Visual Difference Diagnosis: Identifies concrete layout, spacing, typography, color, and hierarchy mismatches and converts pixel-level diagnostics into actionable edits.
  • Integration-Friendly Thresholds: Supports a pass threshold and iterative loop guidance so CI pipelines and design agents can retry until visual acceptance criteria are met.
  • Use Case: Validate a newly generated UI screenshot against approved mocks to decide whether to accept, revise, or fail automatically before further edits.

Quick Start

Use the visual-verdict skill to compare generated_screenshot.png with reference_images ref1.png and ref2.png and return the strict 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 automate visual QA for UI screenshots against reference images?

Automate visual QA by comparing generated UI screenshots to reference images and producing a deterministic pass/fail verdict. This process removes guesswork from visual review cycles and enables automated edit iterations.

What is pixel diff testing for UI design automation?

Pixel diff testing for UI design automation identifies concrete layout, spacing, typography, color, and hierarchy mismatches. It converts pixel-level diagnostics into actionable edits for iterative screenshot-based regression testing.

How do I get a strict JSON verdict for visual regression testing?

Obtain a strict JSON verdict for visual regression testing by running a comparison between generated screenshots and reference images. The output includes a 0-100 score, verdict, category_match flag, differences, suggestions, and reasoning.

Can I integrate visual screenshot comparison into CI pipelines with pass thresholds?

Yes, you can integrate visual screenshot comparison into CI pipelines using pass thresholds and iterative loop guidance. This allows design agents and automated pipelines to retry until visual acceptance criteria are met.

Does visual regression testing work without external dependencies?

Visual regression testing can run without external dependencies. This Skill directly compares UI screenshots to reference images and outputs a compact JSON report suitable for automated pipelines.

What are the limitations of deterministic visual QA verdicts?

Deterministic visual QA verdicts rely on pixel diff comparisons and may require iterative loops to meet acceptance thresholds. It is specifically designed for UI generation and screenshot-based regression testing workflows.