visual-verdict

Compares UI screenshots to references and outputs a structured JSON verdict with similarity scoring.

1|Updated Mar 15, 2026

Search Tags:#design #qa #ui-testing #image-diff #visual-verdict #json-verdict

Ready to run

npx skills add https://github.com/Linnanli/xClaw --skill visual-verdict-linnanli

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: visual-verdict
Source: https://github.com/Linnanli/xClaw/tree/main/.codex/skills/visual-verdict
Command: npx skills add https://github.com/Linnanli/xClaw --skill visual-verdict-linnanli

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill compares generated UI screenshots against one or more reference images and returns a strict JSON verdict that can drive the next edit iteration.

Core Features & Use Cases

Deterministic JSON verdicts for UI screenshot comparisons
Supports multiple reference images and a generated screenshot
Optional category_hint to label UI category

Quick Start

Provide the generated_screenshot and reference_images to receive a deterministic 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 UI visual regression testing and get a JSON verdict for design QA?▼

To perform UI visual regression testing, provide a generated screenshot and reference images to receive a deterministic JSON verdict. This JSON output rates visual similarity and identifies fidelity issues across layouts, typography, and components for design QA.

What is a JSON verdict for image diff comparison and how does it work?▼

A JSON verdict for image diff comparison is a structured output that rates similarity between a generated UI screenshot and reference images. It works by analyzing visual consistency and returning fields like score, verdict, differences, and suggestions to guide edits.

Can I use multiple reference images for UI screenshot comparison?▼

Yes, you can use multiple reference images for UI screenshot comparison. The visual verdict process accepts one or more reference images alongside a generated screenshot to evaluate visual fidelity and produce a strict JSON schema output.

What fields are included in the strict JSON schema for visual verdict output?▼

The strict JSON schema for visual verdict output includes fields such as score, verdict, category_match, differences, suggestions, and reasoning. These fields provide a deterministic evaluation of UI fidelity issues for design QA.

Does visual verdict testing work for validating precise typography and component consistency?▼

Visual verdict testing works for validating precise visual consistency across layouts, typography, and components. It compares generated screenshots against reference images to detect UI fidelity issues and returns a structured JSON verdict.

How do I fix UI fidelity issues identified during design QA?▼

To fix UI fidelity issues identified during design QA, use the suggestions and reasoning fields provided in the JSON verdict output. These fields detail the detected differences between the generated screenshot and reference images to guide the next edit iteration.