Token导航 LogoToken导航

Agent Skill · 研究检索

ux-antipatterns(用户体验反模式)

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。

ux-antipatterns 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,权限为操作浏览器;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 58,594。

总安装 412周安装 17GitHub Stars 12下载 135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:ux-antipatterns(用户体验反模式)
来源仓库:https://github.com/cassiozen/ux-antipatterns
仓库路径:skills/ux-antipatterns
安装命令:
npx skills add https://github.com/cassiozen/ux-antipatterns --skill ux-antipatterns
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/cassiozen/ux-antipatterns --skill ux-antipatterns

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。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。

适用于研究检索类任务中的反模式识别。需结合具体案例理解并规避常见 UX 陷阱。ux-antipatterns 属于研究检索类 Skill,可作为该场景下的辅助能力补充。

来源仓库内路径:skills/ux-antipatterns

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/cassiozen/ux-antipatterns --skill ux-antipatte…;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

    它需要操作浏览器,先在测试目录或测试账号里跑,确认没问题再放到正式项目。

常见问题

6 条

ux-antipatterns 怎么安装?

在终端执行 npx skills add https://github.com/cassiozen/ux-antipatterns --skill ux-antipatterns。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

ux-antipatterns 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。

ux-antipatterns 需要什么权限?

权限级别是操作浏览器。该 Skill 可能涉及浏览器控制能力,使用时可能读取或操作网页内容,需要在受控环境中确认权限边界。

ux-antipatterns 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。

ux-antipatterns 用的人多吗?

公开数据是累计安装 412 次,近一周安装 17 次,下载 135 次,GitHub 12 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。

ux-antipatterns 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。

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。

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
ux-antipatterns

用户体验反模式 · 本页

41212操作浏览器未明确通过

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

UX Anti-Pattern Detection

Scan frontend code for patterns that cause user frustration.

Core Axioms

Before checking individual rules, internalize these. They are the "why" behind every item below.

#AxiomOne-liner
1Acknowledge every actionEvery user action must produce visible feedback within 100ms, even if the result takes seconds.
2Never destroy user inputNot on error, not on navigation, not on timeout, not on refresh.
3State survives the unexpectedRefresh, double-clicks or double submits, network loss — code must handle edge cases.
4Most recent intent winsStale responses must never overwrite a newer user action.
5Explain every constraintIf it's disabled, say why. If it failed, say how to fix it. If it succeeded, say what happened.
6Don't fight the platformBrowser conventions, OS gestures, native controls, and accessibility APIs encode billions of hours of UX research.

When NOT to Use

  • Backend-only code with no UI layer
  • CLI tools or non-visual interfaces
  • Design system tokens/docs without implementation code
  • Pure API or data-layer reviews
  • Performance profiling (unless it manifests as a UX symptom like layout shift)

Workflow

  1. Read references/antipatterns.md to load the full detection heuristics.
  2. Scan the code under review against each applicable anti-pattern category.
  3. Report findings grouped by anti-pattern, citing specific file:line locations.
  4. For each finding, state: the anti-pattern name, the user harm, and a concrete fix.
  5. If no anti-patterns are found, state that the code is clean rather than manufacturing findings.

Anti-Pattern Categories

#CategoryUser Harm
1Layout StabilityClick target moves; wrong thing clicked.
2Feedback & ResponsivenessAction feels ignored; user retries, waits, or loses trust that the system is working.
3Error Handling & RecoveryUser is stuck with no way forward; input destroyed; problem unsolvable without guessing.
4Forms & Input InterferencePlatform fights the user's typing; data mangled, basic editing broken.
5FocusUser is typing and the UI yanks them elsewhere.
6Notifications, Interruptions & DialogsUser's flow broken; attention taxed by noise; forced to parse ambiguous choices under pressure.
7Navigation, Routing & State PersistenceUser can't go back; context evaporates on refresh or redirect.
8Scroll & ViewportContent unreachable or unstable; user fights the interface to see what they came for.
9Timing, Debounce & Race ConditionsActions fire twice, responses arrive stale, sessions expire mid-task; system behaves unpredictably under normal use.
10Accessibility as UXEntire interaction modes broken — keyboard users can't navigate, touch users locked out.
11Visual Layering & RenderingUI elements overlap, clip, or hide each other; controls become unreachable.
12Mobile & Viewport-SpecificKeyboard covers input, layout jumps on scroll, tap targets unresponsive; basic mobile interaction degraded.
13Cumulative Decay & Long-Term UXApp degrades over time; preferences lost, performance rots, stale experiments create inconsistencies.

Quick Reference: Symptom → Category

User complaint / code smellCategory
"Button does nothing when I click it"2. Feedback & Responsiveness
"I clicked the wrong thing — it moved"1. Layout Stability
"I lost my form data"4. Forms & Input Interference
"It says 'Something went wrong' with no explanation"3. Error Handling & Recovery
"The page jumped while I was typing"5. Focus
"I got the same notification 5 times"6. Notifications & Dialogs
"I logged in and it forgot where I was going"7. Navigation & State Persistence
"I scrolled back and lost my place"8. Scroll & Viewport
"My order was placed twice"9. Timing & Race Conditions
"I was filling out a form and it logged me out"9. Timing & Race Conditions
"I clicked delete and it just... deleted it"6. Notifications & Dialogs
"It's been loading for 2 minutes with no progress bar"2. Feedback & Responsiveness
"I can't use this with my keyboard"10. Accessibility as UX
"The dropdown is hidden behind the modal"11. Visual Layering
"The keyboard covers the input on my phone"12. Mobile & Viewport-Specific
"The app gets slower over time"13. Cumulative Decay

Common Mistakes

  • Flagging style preferences as anti-patterns. A non-standard button shape is a design choice, not a UX violation. Only flag patterns that cause measurable user harm per the axioms.
  • Ignoring context. A disabled button inside a wizard step IS explained by the wizard's own flow. Check for nearby explanatory elements before reporting.
  • Suggesting fixes that break accessibility. A fix that adds a visual indicator but removes keyboard access trades one violation for another. Verify fixes against Axiom 6.
  • Over-reporting on handled edge cases. If the code already has an AbortController, don't flag it for race conditions. Read the implementation before reporting.
  • Reporting framework internals as violations. React's key prop remounts, Next.js loading states, or SvelteKit form actions may handle anti-patterns at the framework level. Understand the framework before flagging.

平台分布

Codex

35.66%
按下载量换算48

Claude

30.74%
按下载量换算41

Cursor

17.9%
按下载量换算24

Gemini CLI

10.38%
按下载量换算14

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

通过

权限和风险

操作浏览器

该 Skill 可能涉及浏览器控制能力,使用时可能读取或操作网页内容,需要在受控环境中确认权限边界。

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