ai-visual-accuracy-check

Compare rendered HTML against original PDF pages and produce a similarity score.

Updated Nov 13, 2025
One-click install
npx skills add https://github.com/AbeJitsu/Game-Settings-Panel --skill ai-visual-accuracy-check
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: ai-visual-accuracy-check
Source: https://github.com/AbeJitsu/Game-Settings-Panel/tree/main/.claude/skills/calypso/ai-visual-accuracy-check
Command: npx skills add https://github.com/AbeJitsu/Game-Settings-Panel --skill ai-visual-accuracy-check

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Ensuring generated HTML visually matches its original PDF source is critical but challenging with traditional pixel-diff tools. This Skill uses AI to perform a contextual visual accuracy check, understanding layout intent rather than just pixel differences, saving you manual QA time.

Core Features & Use Cases

  • Contextual AI Comparison: Compares rendered HTML to the original PDF, judging layout, hierarchy, and readability like a human QA reviewer.
  • Objective Scoring: Provides a similarity score (0-100%) and a breakdown by criteria (layout, hierarchy, positioning, typography) for transparent assessment.
  • Blocking Quality Gate: Automatically stops the pipeline if the score falls below a defined threshold (default 85%), preventing visually inaccurate content from proceeding.
  • Use Case: After converting a PDF chapter to HTML, use this Skill to automatically verify that the web version visually aligns with the original PDF, ensuring a consistent and high-fidelity user experience.

Quick Start

Check the visual accuracy of the attached 'chapter_02.html' against the original PDF page '02_page_16.png'.

Frequently Asked Questions about ai-visual-accuracy-check

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

FAQPage Schema
How do I validate that rendered HTML matches the original PDF visually?

Visual validation of HTML against PDF compares layout, hierarchy, positioning, and typography using AI-based contextual analysis rather than pixel-by-pixel matching. This Skill assesses rendered HTML against the original PDF page and produces a similarity score (0–100%) with criterion-level breakdowns to determine if the web version aligns with the source document.

Can I automate QA checks to prevent visually inaccurate HTML from moving through my pipeline?

Yes. This Skill automatically blocks the pipeline when visual similarity falls below a defined threshold (default 85%), stopping inaccurate content from proceeding. It produces a pass/fail recommendation alongside detailed scoring, enabling deterministic gating in content conversion workflows.

What's the difference between AI visual validation and traditional pixel-diff tools for HTML-to-PDF comparison?

Pixel-diff tools flag every minor rendering difference; AI-based visual validation understands layout intent, hierarchy, and readability like a human QA reviewer, reducing false positives and manual review time while providing transparent, criterion-based scoring that explains why content passed or failed.

When should I use visual accuracy checks in my HTML content pipeline?

Use this after converting PDF content to HTML—especially chapters, documents, or reports where visual fidelity to the source is critical. It ensures consistent user experience and prevents layout regressions from reaching production without manual QA overhead.

What output do I get from a visual accuracy assessment?

The Skill returns a numeric similarity score, criterion-level breakdown (layout, hierarchy, positioning, typography), a list of differences noted, and a pass/fail recommendation based on your threshold. This transparent assessment supports both automated gating and informed manual review decisions.