SEO审计工具

# Clone or extract the project
cd seo-audit-mcp

# Install dependencies
npm install

# Install Playwright browsers
npx playwright install chromium

# Install Lighthouse globally (optional but recommended)
npm install -g lighthouse

# Build the project
npm run build

{
  "mcpServers": {
    "seo-audit": {
      "command": "node",
      "args": ["/path/to/seo-audit-mcp/dist/index.js"]
    }
  }
}

# Using npx (after npm link)
claude --mcp-server "node /path/to/seo-audit-mcp/dist/index.js"

Input:
- url (required): The URL to analyze
- waitForSelector: CSS selector to wait for (for JS-heavy pages)
- timeout: Timeout in milliseconds (default: 30000)
- device: 'desktop' or 'mobile' (default: desktop)

Output:
- Meta tags, headings, structured data
- JobPosting schema validation
- JavaScript rendering analysis
- Link and image analysis

Input:
- startUrl (required): Starting URL
- maxPages: Maximum pages to crawl (default: 50)
- maxDepth: Maximum link depth (default: 5)
- includePatterns: Regex patterns to include
- excludePatterns: Regex patterns to exclude

Output:
- Aggregated statistics
- Page type classification (job detail, category, location pages)
- Duplicate detection
- Critical issues and warnings

Input:
- url (required): URL to audit
- device: 'mobile' or 'desktop' (default: mobile)
- categories: Array of categories to audit
- saveReport: Save HTML report (default: false)

Output:
- Performance, Accessibility, Best Practices, SEO scores
- Core Web Vitals (LCP, CLS, TBT, FCP, TTFB)
- Optimization opportunities
- Diagnostics

Input:
- baseUrl (required): Base URL of the site
- includeSitemapUrls: Include full URL list (default: true)
- maxUrls: Max URLs per sitemap (default: 1000)

Output:
- robots.txt rules and issues
- Discovered sitemaps
- Job URL detection
- Recommendations

Input:
- urls (required): Array of URLs to check
- timeout: Timeout per URL in milliseconds

Output:
- Status code, redirect destination, response time per URL

"Analyze the SEO of https://example.com/jobs/software-engineer"

"Crawl https://example.com and analyze their job board SEO. 
Focus on structured data and landing pages."

"Run a Lighthouse audit on https://example.com/jobs for mobile devices"

"Analyze the sitemap for https://example.com and find their job posting pages"

# Run in development mode
npm run dev

# Run tests
npm test

# Lint code
npm run lint

src/
├── index.ts          # Entry point
├── server.ts         # MCP server implementation
├── tools/
│   ├── index.ts      # Tool registry
│   ├── crawl-page.ts # Single page analysis
│   ├── crawl-site.ts # Multi-page crawler
│   ├── lighthouse.ts # Performance audits
│   └── sitemap.ts    # Sitemap/robots analysis
├── types/
│   └── index.ts      # TypeScript definitions
└── utils/
    ├── browser.ts    # Playwright helpers
    └── http.ts       # HTTP utilities

# Reinstall browsers
npx playwright install chromium --force

# On Linux, you may need system dependencies
npx playwright install-deps

# Install globally
npm install -g lighthouse

# Or use npx (slower)
npx lighthouse --version