SEO Analyzer Skill
基于 Google 官方文档的 SEO 自动检测工具。自动分析网址的技术 SEO、内容元数据、性能体验和链接结构，输出符合 Google 最佳实践的检测报告。
使用方法
自动模式 (推荐)
只需告诉 agent 检查某个 URL，skill 会自动完成所有工作：
python skills/seo-analyzer/scripts/seo_analyzer.py --auto <url>Skill 会自动:
- 检查 agent-browser 是否已安装
- 如果已安装，调用 agent-browser 获取页面内容
- 执行完整的 SEO 检查
- 生成详细的检测报告
手动模式
如果 agent-browser 未安装，可以使用手动模式：
# 方式1: 从文件分析
python skills/seo-analyzer/scripts/seo_analyzer.py page.html <url>
# 方式2: 从 stdin 读取 (需要先手动获取 HTML)
agent-browser get source > page.html
python seo_analyzer.py page.html <url>对于 Agent 的调用方式
当用户请求分析某个 URL 时：
# 使用自动模式
python skills/seo-analyzer/scripts/seo_analyzer.py --auto https://example.comAgent 内部会：
- 调用
check_agent_browser_available()检查 agent-browser - 如果可用，调用
fetch_page_with_agent_browser(url)获取页面 - 如果不可用，提示用户安装 agent-browser
- 调用
analyze_page()执行 SEO 分析 - 输出格式化报告
Alternative - save to file first:
# Step 1: Open page
agent-browser open <url> --timeout 30000
# Step 2: Save HTML to file
agent-browser get source > /tmp/page.html
# Step 3: Run analysis
python skills/seo-analyzer/scripts/seo_analyzer.py /tmp/page.html <url>Command-line Options
# Analyze HTML file
python skills/seo-analyzer/scripts/seo_analyzer.py page.html https://example.com
# Read from stdin (pipe from agent-browser)
agent-browser get source | python skills/seo_analyzer.py - https://example.com
# Output in JSON format (for programmatic use)
python seo_analyzer.py page.html https://example.com --jsonWorkflow
- Crawl the URL - Use agent-browser to load the page and capture HTML
- Extract page content - Use
agent-browser get sourceto get HTML - Run SEO checks - Pass HTML to seo_analyzer.py for evaluation
- Generate report - Output findings with severity levels and recommendations
For Agents: How to Analyze a URL
When asked to analyze a website's SEO, follow these steps:
# 1. Open the URL with agent-browser
agent-browser open <target_url> --timeout 30000
# 2. Get the HTML source from the browser
agent-browser get source > /tmp/target_page.html
# 3. Run the SEO analyzer
python skills/seo-analyzer/scripts/seo_analyzer.py /tmp/target_page.html <target_url>Or as a single command:
agent-browser open <target_url> --timeout 30000 && agent-browser get source | python skills/seo-analyzer/scripts/seo_analyzer.py - <target_url>agent-browser open --timeout 30000
### Step 2: Extract the page HTML
After the page loads, get the HTML content:agent-browser get source
Or use get text to extract visible content:agent-browser get text "body"
### Step 3: Save HTML to file and run analysis
Write the HTML to a file, then run the analyzer:Save HTML content to a file
echo "<html>...</html>" > /tmp/page.html
Run the SEO analyzer
python skills/seo-analyzer/scripts/seo_analyzer.py /tmp/page.html <url>
### Integrated workflow (recommended)
Use stdin to pass HTML directly:
agent-browser get source | python skills/seo-analyzer/scripts/seo_analyzer.py - <url>
## Alternative: Analyze existing HTML file
If you already have HTML content saved:
python skills/seo-analyzer/scripts/seo_analyzer.py /path/to/page.html https://example.com/page
## Check Categories
### 1. Technical Requirements (搜索要素/技术要求)
- HTTP status code (must be 200)
- Googlebot accessibility
- Indexable content presence
- HTTPS usage
- robots.txt blocking detection
- noindex directive check
### 2. Title Tag (标题链接)
- Presence of `<title>` element
- Uniqueness across pages
- Descriptive, non-generic text
- Avoid keyword stuffing
- Proper length (not truncated)
- Brand inclusion if appropriate
### 3. Meta Description (片段)
- Presence of `<meta name="description">`
- Unique per page
- Descriptive, summarizes content
- Not keyword-stuffed
- Appropriate length
### 4. Headings Structure (SEO新手指南)
- H1 presence and uniqueness
- Logical heading hierarchy (H1→H2→H3)
- Heading text is descriptive
- Content is organized with headings
### 5. Link Accessibility (链接)
- Links use `<a href="...">` format
- Anchor text is descriptive
- No empty links
- No excessive link clustering
- External links use appropriate rel attributes (nofollow, sponsored, ugc)
### 6. Image Optimization (图像)
- Images have alt attributes
- Alt text is descriptive
- Images are near relevant content
### 7. Structured Data (结构化数据)
- JSON-LD, Microdata, or RDFa presence
- Valid schema.org types
- Required properties for supported types
- Content is visible to users
- Not blocked by robots.txt or noindex
### 8. URL Structure (网址结构)
- Descriptive URLs (not random IDs)
- Human-readable
- Logical directory structure
- Uses hyphens, not underscores
### 9. Canonical URL (规范化)
- Canonical tag presence
- Points to preferred URL
- HTTPS preferred over HTTP
- Consistent with actual URL
### 10. Content Quality (创建实用可靠以用户为中心的内容)
- Unique, original content
- Well-written, readable text
- No spelling/grammar errors
- Content is current/updated
- Provides value to users
### 11. Mobile-Friendliness (移动网站)
- Mobile-compatible design
- Touch-friendly elements
- Readable without zooming
### 12. Core Web Vitals (Core Web Vitals)
- LCP < 2.5s
- INP < 200ms
- CLS < 0.1 (Note: Requires JavaScript runtime measurement)
## Report Format
SEO Analysis Report
URL: <analyzed url> Timestamp: <analysis time>
Summary
- Total Issues: <count>
- Critical: <count>
- Warnings: <count>
- Passed: <count>
Critical Issues
- [Issue name]
- Description - Location: <HTML element/selector> - Recommendation - Reference: seo-docs/<category>/<file>.md
Warnings
- [Issue name]
- ...
Passed Checks
- [Check name]
## Recommendations Priority
1. **Critical**: Must fix (blocks indexing/ranking)
2. **Warning**: Should fix (impacts performance)
3. **Info**: Nice to have (improves appearance)
## Reference Documents
Key seo-docs for SEO analysis:
- `seo基础知识/seo新手指南.md` - General SEO best practices
- `搜索要素/技术要求.md` - Technical requirements
- `排名和搜索结果呈现/标题链接.md` - Title tag guidelines
- `排名和搜索结果呈现/片段.md` - Meta description guidelines
- `排名和搜索结果呈现/结构化数据/了解结构化数据的工作方式.md` - Structured data
- `抓取和编入索引/链接.md` - Link accessibility
- `抓取和编入索引/规范化/如何使用relcanonical.md` - Canonical URLs
- `抓取和编入索引/robots.txt/概览.md` - robots.txt rules
- `排名和搜索结果呈现/网页体验/Core Web Vitals.md` - Core Web Vitals
## Limitations
- Cannot measure Core Web Vitals without JavaScript runtime
- Cannot verify if page is actually indexed in Google
- Cannot access server configuration (SSL, redirects)
- Content quality assessment is heuristic-based
- Mobile-friendliness is approximate without device simulation
## Integration with agent-browser
When analyzing a page:
1. Open page: `agent-browser open <url> --timeout 30000`
2. Get HTML: Use browser's get source function
3. Parse and analyze using SEO rules from seo-docs
4. Report findings with specific element selectors for fixes