Token导航 LogoToken导航

Agent Skill · 研究检索

ui-cloner-quality-check(ui 克隆器质量检查)

用于辅助界面设计、视觉规范、排版、配色、布局和交互体验优化。它适合让 Agent 根据产品场景整理页面结构、生成 UI 方案、检查视觉一致性或改进组件层级。使用时需要结合现有品牌、设计系统和用户任务,不应只堆装饰元素;涉及真实页面改动时,应通过截图或浏览器预览检查文本溢出、对齐和响应式表现。

ui-cloner-quality-check 是研究检索类 Agent Skill,可用宿主为 Codex、Claude、Cursor、Gemini CLI,通过GitHub 仓库获取,权限为只读;许可证信息还不明确,第三方扫描已通过,按安装量在研究检索类的 73,675 个 Skill 里排第 64,068。

总安装 306周安装 13GitHub Stars 91下载 107核验 2026-05-01

安装说明

本站只整理中文说明,不托管安装包,也不代用户安装。
GitHub
许可证
unknown

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:ui-cloner-quality-check(ui 克隆器质量检查)
来源仓库:https://github.com/ragnar-pwninskjold/tech-snacks
仓库路径:skills/ui-cloner-quality-check
安装命令:
npx skills add https://github.com/ragnar-pwninskjold/tech-snacks --skill ui-cloner-quality-check
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 npx skills 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

skills.shnpx skills
npx skills add https://github.com/ragnar-pwninskjold/tech-snacks --skill ui-cloner-quality-check

这个 Skill 是做什么的

研究检索类 Skill 主要解决「找资料、核对事实、把来源整理清楚」的问题,通常会让助手先检索再总结。

用于辅助界面设计、视觉规范、排版、配色、布局和交互体验优化。适合根据产品场景整理页面结构、生成 UI 方案或检查视觉一致性。使用时需结合品牌和设计系统,避免堆砌装饰元素;涉及页面改动应通过截图或浏览器预览检查表现。安装方式:通过 npx skills add 命令从指定 GitHub 仓库添加。

注意:需确认宿主环境支持 Codex、Claude、Cursor 或 Gemini CLI。

来源仓库内路径:skills/ui-cloner-quality-check

上手步骤

4 步
  1. 1

    确认自己的宿主工具

    公开资料标注支持 Codex、Claude、Cursor、Gemini CLI;先确认自己用的是其中之一,再按对应工具的 Skills 目录放置。

  2. 2

    拿到安装命令

    可以直接复制的命令是 npx skills add https://github.com/ragnar-pwninskjold/tech-snacks --skill ui-clon…;命令里的占位符换成自己的路径或密钥再执行。

  3. 3

    安装后重启宿主

    命令执行完要完全退出宿主工具再打开,让新装的 Skill 被重新扫描到,否则列表里可能看不到。

  4. 4

    按最小权限试用

    它标注为只读,先用只读任务试一次,确认行为符合预期再扩大使用范围。

常见问题

6 条

ui-cloner-quality-check 怎么安装?

在终端执行 npx skills add https://github.com/ragnar-pwninskjold/tech-snacks --skill ui-cloner-quality…。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

ui-cloner-quality-check 支持哪些 AI 工具?

公开资料标注支持 Codex、Claude、Cursor、Gemini CLI。其他工具只要能读取本地 Skill 目录,一般也能手动放进去,但要自己确认触发方式是否一致。

ui-cloner-quality-check 需要什么权限?

权限级别是只读。该 Skill 主要提供规则、说明或参考内容,本身偏只读;真正读写文件、联网或执行命令仍取决于宿主 Agent 的任务。

ui-cloner-quality-check 是免费的吗?

公开资料没有标注明确许可证,收费和商用限制都无法确认。正式使用前请到来源仓库核对授权说明。

ui-cloner-quality-check 用的人多吗?

公开数据是累计安装 306 次,近一周安装 13 次,下载 107 次,GitHub 91 星。这些数字来自第三方目录统计,只能当作热度参考,不代表质量或安全性。

ui-cloner-quality-check 安全吗?

第三方扫描结果为通过,但这只覆盖已知的静态检查项,不能替代自己读一遍源码和安装命令。

值得选的地方和要先想清楚的地方

值得选的地方

  • 同时适配 4 种工具,换客户端不用重新找替代品。
  • 第三方扫描已通过,常见的高风险行为没有命中。
  • 标注为只读权限,试用阶段不容易误改本地文件。

先想清楚的地方

  • 许可证没有明确标注,能不能商用、能不能再分发都不确定。
  • 第三方目录整理,版本更新和行为细节以来源仓库为准。

适不适合你

更适合

  • 已经在用 Codex,想直接扩展它能力的人。
  • 需要处理研究检索类任务、又不想每次都手写提示词的人。
  • 只想先试试效果、不希望它改动本地文件的人。

不太适合

  • 需要明确授权才能用到公司项目里的场景。
  • 需要厂商级技术支持和 SLA 的生产系统。

研究检索类 Skill 对比

6 个同类
Skill总安装GitHub 星权限许可证扫描
find-skills

寻找技能 · Codex、Claude

31,199,81316,486操作浏览器MIT有提醒
self-improving-agent

自我改进 Agent · OpenClaw

16,464,0023,396接触敏感数据MIT-0通过
soultrace

灵魂追踪 · Codex、Claude

7,588,94412需要联网未明确有提醒
self-improving

自我完善 · OpenClaw

6,829,2841,046需要联网MIT-0通过
microsoft-foundry

微软代工厂 · Codex、Claude

6,614,144764调用外部服务未明确有提醒
ui-cloner-quality-check

ui 克隆器质量检查 · 本页

30691只读未明确通过

按安装量取同分类前几名;数字来自第三方目录统计,只作热度参考,不代表质量结论。

SKILL.md 原文

UI Cloner — Phase 4: Quality Self-Check

Overview

Before delivering the final prompt, run through the full fidelity checklist below. If any item is missing, generate the missing content and insert it — do not deliver an incomplete prompt.

Announce: "Running Phase 4 — Quality Self-Check before delivery."

Read AUDIT_MODE from plans/01-site-dna.md. High-Fidelity mode must pass all 5 fidelity domains. Standard mode must pass the Core Checklist.


Core Checklist (Both Modes)

Verify the generated prompt contains ALL of the following:

  • Named aesthetic identity — 5–8 word poetic description of the site's soul (derived from user's 3 adjectives + reference site's motion philosophy)
  • Complete design system — named colors (semantic name + descriptive word + hex), typography roles with usage rules, texture/radius/shadow spec
  • Every section has a poetic component name — using the [Evocative Adjective/Noun] + [Functional Description] formula
  • Every interactive component has implementation-level behavior spec — state logic, triggers, resets, JS patterns, CSS property targets
  • At least one cubic-bezier or timing value per animated component — no animation described without exact easing or duration
  • Color hierarchy preserved — reference Primary/Accent/Background/Dark mapped to user's palette in structural roles (not randomly reassigned)
  • Technical requirements section explicitly specified — stack, animation library, lifecycle management, hover implementation, font loading, image sources
  • Execution directive as the final line — italicized or quoted philosophical sentence about HOW to feel while building (not what to build)

High-Fidelity Fidelity Domains (HF Mode Only)

LAYOUT FIDELITY

  • ASCII wireframe present for EVERY section (not omitted, not replaced with prose)
  • Internal component layouts documented with dimensions and positioning logic
  • Grid/flex layout system specified per section — the EXACT column structure (not "use a grid")

VISUAL COMPOSITION FIDELITY

  • Hero composition fully specified — if the Site DNA shows a multi-element composition (Composition Map exists), the prompt must specify: element count, arrangement, per-element blur/opacity/rotation, and spatial relationship to centerpiece. A hero described as "phone mockup" when the DNA shows "phone + fanned cards" is a CRITICAL failure.
  • No flattened descriptions — search the prompt for vague descriptions like "shows [device] with [app]", "animation plays", "cards appear", "warm palette". Each must be replaced with exact specifications. Every visual element must be individually described.
  • Opaque media decomposed — any Lottie, SVG, or canvas element is described by its rendered visual output (what you see), not its DOM representation (one node)

ANIMATION FIDELITY

  • Every animated element has a timeline entry with t=Xms notation
  • At least one cubic-bezier() or named easing value per animated component
  • Scroll trigger positions specified — start: top 80% viewport format (not "when visible")
  • Stateful components have full state machine notation

INTERACTION FIDELITY

  • Every interactive element has a DEFAULT → HOVER property diff table
  • Special mechanisms (pseudo-element slides, clip-path reveals) are explicitly named
  • Transition durations and easings specified per interaction

DESIGN SYSTEM FIDELITY

  • Complete palette with semantic names + hex codes
  • Typography scale with all 5+ roles specified with exact values
  • Typographic drama ratio from Step 1.2 ⚑ DRAMA NOTES explicitly preserved

TECHNICAL FIDELITY

  • Exact tech stack specified with library names and versions
  • Animation lifecycle management specified (e.g., gsap.context() / cleanup)
  • Font loading method specified
  • Scroll trigger setup specified
  • Execution directive present as final line

Zero Generic Language Enforcement (Both Modes)

Scan the entire prompt for any instance of the following phrases. Each is a fidelity failure — replace with exact values:

Generic phraseReplace with
"some animation"specific element + t=Xms timeline or duration
"nice hover effect"exact property diff: DEFAULT→HOVER state table
"smooth transition"exact duration + easing value
"add appropriate padding"exact px/rem value
"use a gradient"exact direction + stop values + hex codes
"make it feel premium"specific design decisions that produce that feeling
"add content here"exact content or Unsplash URL with aesthetic description
"similar to"copy the exact value, don't reference by comparison

Additional banned flattening patterns:

  • "Shows [device] with [app/interface/UI]" → must describe what's ON the screen AND what's around the device
  • "Large [animation/illustration/graphic]" → must specify dimensions, element count, arrangement
  • "Warm/cool palette" → must list every color with hex and role
  • "Text animates in" → must specify per-word/per-line, stagger, easing, duration
  • "Cards/elements appear" → must specify how many, from where, in what order, with what timing
  • "[N] items" without describing layout → must specify grid/flex/stack, gaps, alignment

Failure Protocol

If any item is unchecked:

  1. Identify exactly what's missing
  2. Generate the missing content inline using the correct format from the Site DNA (don't soften or approximate)
  3. Insert it into the correct location in the prompt
  4. Re-check that item before proceeding

Do not deliver until all applicable items are checked.


Delivery

Once all applicable items pass:

  1. Save output: Write the final verified prompt to plans/04-final-prompt.md in the current project directory.
  2. Output the complete final prompt in a single fenced code block so the user can copy and paste it cleanly.
  3. Confirm: "Your replication prompt is ready. Paste it into Claude or your preferred code generation tool to build the site. All phase outputs are saved in plans/."

平台分布

Codex

35.56%
按下载量换算38

Claude

29.32%
按下载量换算31

Cursor

21.06%
按下载量换算23

Gemini CLI

9.85%
按下载量换算11

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

通过

权限和风险

只读

该 Skill 主要提供规则、说明或参考内容,本身偏只读;真正读写文件、联网或执行命令仍取决于宿主 Agent 的任务。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。目前只有一个公开来源,建议同时核对源仓库。

继续浏览同类 Skills