visual-verdict

Compare screenshots against reference images and return a strict JSON verdict.

Updated Apr 3, 2026
One-click install
npx skills add https://github.com/pextacy/stellar --skill visual-verdict-pextacy
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: visual-verdict
Source: https://github.com/pextacy/stellar/tree/main/skills/visual-verdict
Command: npx skills add https://github.com/pextacy/stellar --skill visual-verdict-pextacy

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Quickly identify UI discrepancies by comparing a generated screenshot against reference images and produce a strict JSON verdict to drive the next edit iteration.

Core Features & Use Cases

  • Surface pixel-level differences between current UI renders and references.
  • Produce a deterministic verdict JSON containing score, verdict, category_match, differences, suggestions, and reasoning to guide design iterations.
  • Valid across dashboards, apps, and component libraries where layout and typography fidelity matter.

Quick Start

Provide a reference image and the generated screenshot to receive a JSON verdict that guides the next edit.

Frequently Asked Questions about visual-verdict

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

FAQPage Schema
How do I detect visual regression in UI screenshots against a reference image?

To detect visual regression, compare your generated UI screenshot against a reference image to surface pixel-level differences. This process identifies layout and typography discrepancies across dashboards, apps, and component libraries where fidelity matters.

What is a visual verdict in UI testing and how does it guide design iterations?

A visual verdict in UI testing is a strict JSON object containing a score, verdict, category_match, differences, suggestions, and reasoning. It provides deterministic guidance for targeted edits by identifying exact UI discrepancies between current renders and references.

Can I use visual regression testing for component libraries and dashboards?

Yes, visual regression testing is valid across dashboards, apps, and component libraries. It effectively identifies layout and typography fidelity issues by comparing current UI renders against reference images to produce actionable edit guidance.

How do I get a JSON verdict from comparing two UI screenshots?

Provide a reference image and the generated screenshot to receive a deterministic JSON verdict. The output includes a score, verdict, category_match, differences, suggestions, and reasoning to drive targeted edits for your next iteration.

Does pixel-level image diff replace manual visual QA for web apps?

Pixel-level image diff automates visual QA by producing strict JSON verdicts that guide edits, reducing manual review. It is best applied where deterministic guidance and pixel-level fidelity are required across UI renders.

Why does my UI screenshot comparison return a mismatch despite matching layouts?

A visual verdict mismatch can occur due to subtle pixel-level differences in typography or rendering. The returned JSON details specific differences and reasoning to help you determine whether the discrepancy requires a targeted edit.