AGENT SKILL

visual-verdict

Structured visual QA verdict for screenshot-to-reference comparisons

License 许可
GPL-3.0
最近更新
2026-04-24
Stars
★ 1
Forks
3

01 前言 Front Matter

name
visual-verdict
level
2
description
Structured visual QA verdict for screenshot-to-reference comparisons

03 SKILL.md 正文

<Purpose> Use this skill to compare generated UI screenshots against one or more reference images and return a strict JSON verdict that can drive the next edit iteration. </Purpose>

<Use_When>

  • The task includes visual fidelity requirements (layout, spacing, typography, component styling)
  • You have a generated screenshot and at least one reference image
  • You need deterministic pass/fail guidance before continuing edits </Use_When>

<Inputs>

  • reference_images[] (one or more image paths)
  • generated_screenshot (current output image)
  • Optional: category_hint (e.g., hackernews, sns-feed, dashboard) </Inputs>

<Output_Contract> Return JSON only with this exact shape:

{
  "score": 0,
  "verdict": "revise",
  "category_match": false,
  "differences": ["..."],
  "suggestions": ["..."],
  "reasoning": "short explanation"
}

Rules:

  • score: integer 0-100
  • verdict: short status (pass, revise, or fail)
  • category_match: true when the generated screenshot matches the intended UI category/style
  • differences[]: concrete visual mismatches (layout, spacing, typography, colors, hierarchy)
  • suggestions[]: actionable next edits tied to the differences
  • reasoning: 1-2 sentence summary

<Threshold_And_Loop>

  • Target pass threshold is 90+.
  • If score < 90, continue editing and rerun /startup-harness:visual-verdict before any further visual review pass.
  • Do not treat the visual task as complete until the next screenshot clears the threshold. </Threshold_And_Loop>

<Debug_Visualization> When mismatch diagnosis is hard:

  1. Keep $visual-verdict as the authoritative decision.
  2. Use pixel-level diff tooling (pixel diff / pixelmatch overlay) as a secondary debug aid to localize hotspots.
  3. Convert pixel diff hotspots into concrete differences[] and suggestions[] updates. </Debug_Visualization>

<Example>

{
  "score": 87,
  "verdict": "revise",
  "category_match": true,
  "differences": [
    "Top nav spacing is tighter than reference",
    "Primary button uses smaller font weight"
  ],
  "suggestions": [
    "Increase nav item horizontal padding by 4px",
    "Set primary button font-weight to 600"
  ],
  "reasoning": "Core layout matches, but style details still diverge."
}

</Example>

Task: {{ARGUMENTS}}

04 同仓库更多 Skills

Cheggin
Cheggin/request-for-startups

writer-memory

Agentic memory system for writers - track characters, relationships, scenes, and themes

★ 1
Cheggin
Cheggin/request-for-startups

wiki

LLM Wiki — persistent markdown knowledge base that compounds across sessions (Karpathy model)

★ 1
Cheggin
Cheggin/request-for-startups

website-creation

Build production-quality SaaS websites with opinionated design presets. Use when creating any startup website. The user MUST pick a design style before building. Enforces shadcn/ui, Figma design principles, specific CSS values per style, and anti-AI-writing. Load alongside anti-ai-writing skill. LIGHT MODE ONLY.

设计师内容创作者视觉设计写作与文案
★ 1
Cheggin
Cheggin/request-for-startups

visual-qa-pipeline

Screenshot every page with Playwright, feed to visual QA agent for design evaluation, report results to Slack. Use after every build phase to verify UI matches the chosen design preset. Integrates with the Playwright skill and the website-creation preset system.

设计师邮件与沟通测试与质量
★ 1
Cheggin
Cheggin/request-for-startups

verify

Verify that a change really works before you claim completion

★ 1
Cheggin
Cheggin/request-for-startups

user-feedback-collector

Build an in-app feedback widget that collects user feedback, categorizes it by type (bug, feature request, general), routes it to appropriate agents, and converts feature requests into GitHub Issues. Use when adding feedback collection, building feedback widgets, setting up user feedback routing, or aggregating feedback trends.

★ 1