web-design-reviewer

作者: microsoft

此技能可對本地或遠端運行的網站進行視覺檢查,以識別並修復設計問題。觸發條件為「review website…」等請求。

npx skills add https://github.com/microsoft/githubcopilot_customized --skill web-design-reviewer

Web Design Reviewer

This skill enables visual inspection and validation of website design quality, identifying and fixing issues at the source code level.

Scope of Application

  • Static sites (HTML/CSS/JS)
  • SPA frameworks such as React / Vue / Angular / Svelte
  • Full-stack frameworks such as Next.js / Nuxt / SvelteKit
  • CMS platforms such as WordPress / Drupal
  • Any other web application

Prerequisites

Required

  1. Target website must be running

    • Local development server (e.g., http://localhost:3000)
    • Staging environment
    • Production environment (for read-only reviews)
  2. Browser automation must be available

    • Screenshot capture
    • Page navigation
    • DOM information retrieval
  3. Access to source code (when making fixes)

    • Project must exist within the workspace

Workflow Overview

flowchart TD
    A[Step 1: Information Gathering] --> B[Step 2: Visual Inspection]
    B --> C[Step 3: Issue Fixing]
    C --> D[Step 4: Re-verification]
    D --> E{Issues Remaining?}
    E -->|Yes| B
    E -->|No| F[Completion Report]

Step 1: Information Gathering Phase

1.1 URL Confirmation

If the URL is not provided, ask the user:

Please provide the URL of the website to review (e.g., http://localhost:3000)

1.2 Understanding Project Structure

When making fixes, gather the following information:

ItemExample Question
FrameworkAre you using React / Vue / Next.js, etc.?
Styling MethodCSS / SCSS / Tailwind / CSS-in-JS, etc.
Source LocationWhere are style files and components located?
Review ScopeSpecific pages only or entire site?

1.3 Automatic Project Detection

Attempt automatic detection from files in the workspace:

Detection targets:
├── package.json     → Framework and dependencies
├── tsconfig.json    → TypeScript usage
├── tailwind.config  → Tailwind CSS
├── next.config      → Next.js
├── vite.config      → Vite
├── nuxt.config      → Nuxt
└── src/ or app/     → Source directory

1.4 Identifying Styling Method

MethodDetectionEdit Target
Pure CSS*.css filesGlobal CSS or component CSS
SCSS/Sass*.scss, *.sassSCSS files
CSS Modules*.module.cssModule CSS files
Tailwind CSStailwind.config.*className in components
styled-componentsstyled. in codeJS/TS files
Emotion@emotion/ importsJS/TS files
CSS-in-JS (other)Inline stylesJS/TS files

Step 2: Visual Inspection Phase

2.1 Page Traversal

  1. Navigate to the specified URL
  2. Capture screenshots
  3. Retrieve DOM structure/snapshot (if possible)
  4. If additional pages exist, traverse through navigation

2.2 Inspection Items

Layout Issues

IssueDescriptionSeverity
Element OverflowContent overflows from parent element or viewportHigh
Element OverlapUnintended overlapping of elementsHigh
Alignment IssuesGrid or flex alignment problemsMedium
Inconsistent SpacingPadding/margin inconsistenciesMedium
Text ClippingLong text not handled properlyMedium

Responsive Issues

IssueDescriptionSeverity
Non-mobile FriendlyLayout breaks on small screensHigh
Breakpoint IssuesUnnatural transitions when screen size changesMedium
Touch TargetsButtons too small on mobileMedium

Accessibility Issues

IssueDescriptionSeverity
Insufficient ContrastLow contrast ratio between text and backgroundHigh
No Focus StateCannot determine state during keyboard navigationHigh
Missing alt TextNo alternative text for imagesMedium

Visual Consistency

IssueDescriptionSeverity
Font InconsistencyMixed font familiesMedium
Color InconsistencyNon-unified brand colorsMedium
Spacing InconsistencyNon-uniform spacing between similar elementsLow

2.3 Viewport Testing (Responsive)

Test at the following viewports:

NameWidthRepresentative Device
Mobile375pxiPhone SE/12 mini
Tablet768pxiPad
Desktop1280pxStandard PC
Wide1920pxLarge display

Step 3: Issue Fixing Phase

3.1 Issue Prioritization

block-beta
    columns 1
    block:priority["Priority Matrix"]
        P1["P1: Fix Immediately\n(Layout issues affecting functionality)"]
        P2["P2: Fix Next\n(Visual issues degrading UX)"]
        P3["P3: Fix If Possible\n(Minor visual inconsistencies)"]
    end

3.2 Identifying Source Files

Identify source files from problematic elements:

  1. Selector-based Search

    • Search codebase by class name or ID
    • Explore style definitions with grep_search
  2. Component-based Search

    • Identify components from element text or structure
    • Explore related files with semantic_search
  3. File Pattern Filtering

    Style files: src/**/*.css, styles/**/*
    Components: src/components/**/*
    Pages: src/pages/**, app/**
    

3.3 Applying Fixes

Framework-specific Fix Guidelines

See references/framework-fixes.md for details.

Fix Principles

  1. Minimal Changes: Only make the minimum changes necessary to resolve the issue
  2. Respect Existing Patterns: Follow existing code style in the project
  3. Avoid Breaking Changes: Be careful not to affect other areas
  4. Add Comments: Add comments to explain the reason for fixes where appropriate

Step 4: Re-verification Phase

4.1 Post-fix Confirmation

  1. Reload browser (or wait for development server HMR)
  2. Capture screenshots of fixed areas
  3. Compare before and after

4.2 Regression Testing

  • Verify that fixes haven't affected other areas
  • Confirm responsive display is not broken

4.3 Iteration Decision

flowchart TD
    A{Issues Remaining?}
    A -->|Yes| B[Return to Step 2]
    A -->|No| C[Proceed to Completion Report]

Iteration Limit: If more than 3 fix attempts are needed for a specific issue, consult the user


Output Format

Review Results Report

# Web Design Review Results

## Summary

| Item | Value |
|------|-------|
| Target URL | {URL} |
| Framework | {Detected framework} |
| Styling | {CSS / Tailwind / etc.} |
| Tested Viewports | Desktop, Mobile |
| Issues Detected | {N} |
| Issues Fixed | {M} |

## Detected Issues

### [P1] {Issue Title}

- **Page**: {Page path}
- **Element**: {Selector or description}
- **Issue**: {Detailed description of the issue}
- **Fixed File**: `{File path}`
- **Fix Details**: {Description of changes}
- **Screenshot**: Before/After

### [P2] {Issue Title}
...

## Unfixed Issues (if any)

### {Issue Title}
- **Reason**: {Why it was not fixed/could not be fixed}
- **Recommended Action**: {Recommendations for user}

## Recommendations

- {Suggestions for future improvements}

Required Capabilities

CapabilityDescriptionRequired
Web Page NavigationAccess URLs, page transitions✅
Screenshot CapturePage image capture✅
Image AnalysisVisual issue detection✅
DOM RetrievalPage structure retrievalRecommended
File Read/WriteSource code reading and editingRequired for fixes
Code SearchCode search within projectRequired for fixes

Reference Implementation

Implementation with Playwright MCP

Playwright MCP is recommended as the reference implementation for this skill.

CapabilityPlaywright MCP ToolPurpose
Navigationbrowser_navigateAccess URLs
Snapshotbrowser_snapshotRetrieve DOM structure
Screenshotbrowser_take_screenshotImages for visual inspection
Clickbrowser_clickInteract with interactive elements
Resizebrowser_resizeResponsive testing
Consolebrowser_console_messagesDetect JS errors

Configuration Example (MCP Server)

{
  "mcpServers": {
    "playwright": {
      "command": "npx",
      "args": ["-y", "@playwright/mcp@latest", "--caps=vision"]
    }
  }
}

Other Compatible Browser Automation Tools

ToolFeatures
SeleniumBroad browser support, multi-language support
PuppeteerChrome/Chromium focused, Node.js
CypressEasy integration with E2E testing
WebDriver BiDiStandardized next-generation protocol

The same workflow can be implemented with these tools. As long as they provide the necessary capabilities (navigation, screenshot, DOM retrieval), the choice of tool is flexible.


Best Practices

DO (Recommended)

  • ✅ Always save screenshots before making fixes
  • ✅ Fix one issue at a time and verify each
  • ✅ Follow the project's existing code style
  • ✅ Confirm with user before major changes
  • ✅ Document fix details thoroughly

DON'T (Not Recommended)

  • ❌ Large-scale refactoring without confirmation
  • ❌ Ignoring design systems or brand guidelines
  • ❌ Fixes that ignore performance
  • ❌ Fixing multiple issues at once (difficult to verify)

Troubleshooting

Problem: Style files not found

  1. Check dependencies in package.json
  2. Consider the possibility of CSS-in-JS
  3. Consider CSS generated at build time
  4. Ask user about styling method

Problem: Fixes not reflected

  1. Check if development server HMR is working
  2. Clear browser cache
  3. Rebuild if project requires build
  4. Check CSS specificity issues

Problem: Fixes affecting other areas

  1. Rollback changes
  2. Use more specific selectors
  3. Consider using CSS Modules or scoped styles
  4. Consult user to confirm impact scope

來自 microsoft 的更多技能

oss-growth
microsoft
開源增長駭客角色
agent-framework-azure-ai-py
microsoft
使用Microsoft Agent Framework Python SDK(agent-framework-azure-ai)构建Azure AI Foundry代理。适用于使用AzureAIAgentsProvider创建持久化代理、使用托管工具(代码解释器、文件搜索、网络搜索)、集成MCP服务器、管理对话线程或实现流式响应。涵盖函数工具、结构化输出和多工具代理。
development
airunway-aks-setup
microsoft
在AKS上設定AI Runway——從裸叢集到執行模型。涵蓋叢集驗證、控制器安裝、GPU評估、供應商設定及首次部署。時機:「設定AI Runway」、「上線AKS叢集」、「安裝AI Runway」、「airunway設定」、「部署模型至AKS」、「在AKS上進行GPU推論」、「在AKS上設定KAITO」、「在AKS上執行LLM」、「在AKS上使用vLLM」、「在AKS上設定模型服務」、「AI Runway控制器」。
devops
appinsights-instrumentation
microsoft
使用Azure Application Insights檢測Web應用程式的指南。提供遙測模式、SDK設定與組態參考。適用時機:如何檢測應用程式、App Insights SDK、遙測模式、什麼是App Insights、Application Insights指南、檢測範例、APM最佳實踐。
devops
applicationinsights-web-ts
microsoft
使用Application Insights JavaScript SDK(@microsoft/applicationinsights-web)為瀏覽器/Web應用程式進行檢測。適用於真實使用者監控(RUM)——頁面檢視、點擊、AJAX/fetch依賴、例外、自訂事件,以及與後端OpenTelemetry追蹤關聯的瀏覽器端GenAI代理追蹤。涵蓋SDK載入器指令碼與npm設定、框架擴充(React、React Native、Angular)、點擊分析、遙測初始化器,以及從瀏覽器發出的代理/工具/模型span的OTel GenAI語意慣例。
devops
azure-ai-anomalydetector-java
microsoft
使用適用於 Java 的 Azure AI 異常偵測器 SDK 建置異常偵測應用程式。在實作單變量/多變量異常偵測、時間序列分析或 AI 驅動監控時使用。
development
azure-ai-language-conversations-py
microsoft
使用 azure-ai-language-conversations Python SDK 實作對話語言理解(CLU)。當使用 ConversationAnalysisClient 分析對話意圖與實體、建置 NLP 功能,或將語言理解整合至應用程式時使用。
development
azure-ai-ml-py
microsoft
Azure Machine Learning SDK v2 for Python。用於機器學習工作區、作業、模型、資料集、計算資源與管線。 觸發詞:「azure-ai-ml」、「MLClient」、「workspace」、「model registry」、「training jobs」、「datasets」。
development