Agentic SEO Skill

Skill by ara.so — AI Agent Skills collection.

The Agentic SEO Skill is an LLM-first SEO analysis framework designed for AI coding assistants and agent IDEs. It provides 16 specialized sub-skills, 10 specialist agents, and 89 Python scripts for evidence-based SEO analysis. The skill follows a reasoning-first workflow: collect evidence, analyze with LLM proof, apply confidence labels, prioritize by impact, and produce actionable plans.

由ara.so开发的Skill — AI Agent技能合集。

Agentic SEO Skill是一款面向AI编码助手和Agent IDE的、以LLM为核心的SEO分析框架。它提供16项专业子技能、10个专业Agent和89个Python脚本，用于基于证据的SEO分析。该Skill遵循“推理优先”的工作流程：收集证据、借助LLM进行带依据的分析、添加置信度标签、按影响优先级排序，并生成可执行计划。

Installation

安装

Quick Install (No Cloning Required)

快速安装（无需克隆）

Linux/macOS:

bash

undefined

Linux/macOS:

bash

undefined

Install to all supported IDEs

curl -fsSL https://raw.githubusercontent.com/Bhanunamikaze/Agentic-SEO-Skill/main/install.sh | bash -s -- --online

Install to Claude Code only

curl -fsSL https://raw.githubusercontent.com/Bhanunamikaze/Agentic-SEO-Skill/main/install.sh | bash -s -- --online --target claude

Install to specific project

curl -fsSL https://raw.githubusercontent.com/Bhanunamikaze/Agentic-SEO-Skill/main/install.sh | bash -s -- --online --target all --project-dir /path/to/project


**Windows (PowerShell 7+):**
```powershell
irm https://raw.githubusercontent.com/Bhanunamikaze/Agentic-SEO-Skill/main/install.ps1 -OutFile install.ps1
powershell -ExecutionPolicy Bypass -File .\install.ps1 --online

curl -fsSL https://raw.githubusercontent.com/Bhanunamikaze/Agentic-SEO-Skill/main/install.sh | bash -s -- --online --target all --project-dir /path/to/project


**Windows (PowerShell 7+):**
```powershell
irm https://raw.githubusercontent.com/Bhanunamikaze/Agentic-SEO-Skill/main/install.ps1 -OutFile install.ps1
powershell -ExecutionPolicy Bypass -File .\install.ps1 --online

From Source

从源码安装

bash

git clone https://github.com/Bhanunamikaze/Agentic-SEO-Skill.git
cd Agentic-SEO-Skill

bash

git clone https://github.com/Bhanunamikaze/Agentic-SEO-Skill.git
cd Agentic-SEO-Skill

Install to Claude Code with dependencies

bash install.sh --target claude --install-deps

Install with Playwright for visual analysis

bash install.sh --target claude --install-deps --install-playwright

undefined

bash install.sh --target claude --install-deps --install-playwright

undefined

Installation Targets

安装目标

Target	Location	Format
`claude`	`~/.claude/skills/seo`	Skill directory
`codex`	`~/.codex/skills/seo`	Skill directory
`cursor`	`<project>/.cursor/rules/seo.mdc`	MDC rule
`windsurf`	`<project>/.windsurf/rules/seo.md`	Windsurf rule
`copilot`	`<project>/.github/copilot-instructions.md`	Repo instructions
`cline`	`<project>/.clinerules`	Project rules
`continue`	`<project>/.continue/prompts/seo.prompt`	Slash command
`antigravity`	`<project>/.agent/skills/seo`	Skill directory

目标	位置	格式
`claude`	`~/.claude/skills/seo`	Skill目录
`codex`	`~/.codex/skills/seo`	Skill目录
`cursor`	`<project>/.cursor/rules/seo.mdc`	MDC规则
`windsurf`	`<project>/.windsurf/rules/seo.md`	Windsurf规则
`copilot`	`<project>/.github/copilot-instructions.md`	仓库说明文档
`cline`	`<project>/.clinerules`	项目规则
`continue`	`<project>/.continue/prompts/seo.prompt`	斜杠命令
`antigravity`	`<project>/.agent/skills/seo`	Skill目录

Core Workflow

核心工作流程

The skill follows an LLM-first analysis pattern:

Collect Evidence — Use
```
read_url_content
```
MCP tool or evidence collection scripts
Analyze with LLM — Apply explicit proof and reasoning for each finding
Label Confidence — Mark findings as
```
Confirmed
```
,
```
Likely
```
, or
```
Hypothesis
```
Prioritize — Rank by impact and implementation effort
Generate Action Plan — Produce structured, evidence-backed recommendations

All audits must follow the

resources/references/llm-audit-rubric.md

standard.

该Skill遵循LLM优先的分析模式：

收集证据 — 使用
```
read_url_content
```
MCP工具或证据收集脚本
借助LLM分析 — 为每个发现提供明确的依据和推理过程
标记置信度 — 将发现标记为
```
Confirmed
```
（已确认）、
```
Likely
```
（大概率）或
```
Hypothesis
```
（假设）
优先级排序 — 按影响程度和实施难度排序
生成行动计划 — 生成结构化、基于证据的建议

所有审计必须遵循

resources/references/llm-audit-rubric.md

标准。

16 Specialized Sub-Skills

16项专业子技能

1. SEO Audit (Full Site)

1. SEO审计（全站）

Run comprehensive site-wide audits with evidence-backed scoring:

python

undefined

运行全面的站点审计并生成带证据支撑的评分：

python

undefined

Quick full audit with all outputs

python3 scripts/audit_runner.py https://example.com

Outputs:

- audit_results.json

- audit_report.html

- FULL-AUDIT-REPORT.md

- ACTION-PLAN.md


**Agent prompt:**

Use the seo audit sub-skill to analyze https://example.com — follow the LLM audit rubric, collect evidence first with read_url_content, then reason through technical, content, and performance issues.

undefined


**Agent指令:**

Use the seo audit sub-skill to analyze https://example.com — follow the LLM audit rubric, collect evidence first with read_url_content, then reason through technical, content, and performance issues.

undefined

2. SEO Page (Single Page Analysis)

2. SEO页面分析（单页）

Deep analysis of individual pages:

python

undefined

对单个页面进行深度分析：

python

undefined

Fetch page content

python3 scripts/fetch_page.py https://example.com/page

Parse HTML for SEO signals

python3 scripts/parse_html.py example_com_page.html

Check Core Web Vitals

python3 scripts/pagespeed.py https://example.com/page


**Agent prompt:**

Analyze this single page for SEO: https://example.com/about — check title, meta description, headings hierarchy, internal links, schema markup, and Core Web Vitals.

undefined

python3 scripts/pagespeed.py https://example.com/page


**Agent指令:**

Analyze this single page for SEO: https://example.com/about — check title, meta description, headings hierarchy, internal links, schema markup, and Core Web Vitals.

undefined

3. SEO Technical

3. SEO技术审计

Crawlability, indexability, security, mobile-friendliness, and AI crawler policies:

python

undefined

分析可抓取性、可索引性、安全性、移动端适配及AI爬虫规则：

python

undefined

Check robots.txt policies

python3 scripts/robots_checker.py https://example.com

Run indexability matrix (robots, meta robots, canonicals, status codes)

python3 scripts/indexability_matrix.py https://example.com

Multi-page crawl audit

python3 scripts/crawl_audit.py https://example.com --max-depth 3


**Agent prompt:**

Run a technical SEO audit on https://example.com — check robots.txt, XML sitemaps, HTTPS, mobile responsiveness, JS rendering, and AI crawler access (GPTBot, ClaudeBot).

undefined

python3 scripts/crawl_audit.py https://example.com --max-depth 3


**Agent指令:**

Run a technical SEO audit on https://example.com — check robots.txt, XML sitemaps, HTTPS, mobile responsiveness, JS rendering, and AI crawler access (GPTBot, ClaudeBot).

undefined

4. SEO Content

4. SEO内容审计

Content quality assessment with E-E-A-T framework (September 2025 Quality Rater Guidelines):

python

undefined

基于E-E-A-T框架（2025年9月质量评估指南）评估内容质量：

python

undefined

Extract article content

python3 scripts/fetch_page.py https://example.com/article

Run content quality checks (agent analyzes with LLM)

Check: word count, readability, E-E-A-T signals, topical depth


**Agent prompt:**

Assess content quality for https://example.com/blog/post using E-E-A-T criteria — check expertise signals, author credentials, content depth, readability, and December 2025 core update alignment.

undefined


**Agent指令:**

Assess content quality for https://example.com/blog/post using E-E-A-T criteria — check expertise signals, author credentials, content depth, readability, and December 2025 core update alignment.

undefined

5. SEO Schema

5. SEO Schema分析

Schema.org detection, validation, and JSON-LD generation:

python

undefined

检测、验证Schema.org标记并生成JSON-LD：

python

undefined

Validate existing schema markup

python3 scripts/validate_schema.py example_com.html

Check for deprecated schema types

python3 scripts/parse_html.py example_com.html --extract-schema


**Agent prompt:**

Validate schema markup on https://example.com — check JSON-LD syntax, required fields, deprecated types, and suggest appropriate schema.org types (Article, Product, LocalBusiness, etc.).


**Example JSON-LD generation:**
```python

python3 scripts/parse_html.py example_com.html --extract-schema


**Agent指令:**

Validate schema markup on https://example.com — check JSON-LD syntax, required fields, deprecated types, and suggest appropriate schema.org types (Article, Product, LocalBusiness, etc.).


**JSON-LD生成示例:**
```python

Agent generates schema based on page content

schema = { "@context": "https://schema.org", "@type": "Article", "headline": "Page Title", "author": { "@type": "Person", "name": "Author Name" }, "datePublished": "2026-01-15", "dateModified": "2026-05-27", "image": "https://example.com/image.jpg", "publisher": { "@type": "Organization", "name": "Publisher Name", "logo": { "@type": "ImageObject", "url": "https://example.com/logo.png" } } }

undefined

schema = { "@context": "https://schema.org", "@type": "Article", "headline": "Page Title", "author": { "@type": "Person", "name": "Author Name" }, "datePublished": "2026-01-15", "dateModified": "2026-05-27", "image": "https://example.com/image.jpg", "publisher": { "@type": "Organization", "name": "Publisher Name", "logo": { "@type": "ImageObject", "url": "https://example.com/logo.png" } } }

undefined

6. SEO Sitemap

6. SEO站点地图分析

XML sitemap validation and generation:

python

undefined

验证并生成XML站点地图：

python

undefined

Discover and validate sitemaps

python3 scripts/sitemap_checker.py https://example.com

Check for issues: 404s, lastmod quality, URL limits (50,000 per file)


**Agent prompt:**

Audit XML sitemaps for https://example.com — check discovery (robots.txt reference), URL limits, 404s, lastmod accuracy, and sitemap index structure.

undefined


**Agent指令:**

Audit XML sitemaps for https://example.com — check discovery (robots.txt reference), URL limits, 404s, lastmod accuracy, and sitemap index structure.

undefined

7. SEO Images

7. SEO图片优化审计

Image optimization audit (alt text, formats, lazy loading, CLS):

python

undefined

图片优化审计（替代文本、格式、懒加载、CLS）：

python

undefined

Generate image inventory

python3 scripts/image_inventory.py https://example.com

Outputs: alt text coverage, dimensions, loading behavior, LCP candidates


**Agent prompt:**

Analyze image SEO for https://example.com — check alt text, file formats (WebP), lazy loading, responsive images (srcset), and LCP image optimization.

undefined


**Agent指令:**

Analyze image SEO for https://example.com — check alt text, file formats (WebP), lazy loading, responsive images (srcset), and LCP image optimization.

undefined

8. SEO GEO (Generative Engine Optimization)

8. SEO GEO（生成式引擎优化）

Optimize for AI Overviews, ChatGPT, Perplexity:

Agent prompt:

Optimize this content for AI search engines (ChatGPT, Perplexity, Google AI Overviews) — add clear definitions, structured facts, citations, and FAQ sections.

Pattern:

Use concise definitions (20-30 words)
Add structured lists and tables
Include citations and sources
Create FAQ sections for common queries

针对AI概览、ChatGPT、Perplexity进行优化：

Agent指令:

Optimize this content for AI search engines (ChatGPT, Perplexity, Google AI Overviews) — add clear definitions, structured facts, citations, and FAQ sections.

优化模式:

使用简洁定义（20-30词）
添加结构化列表和表格
包含引用和来源
创建常见问题（FAQ）板块

9. SEO AEO (Answer Engine Optimization)

9. SEO AEO（问答引擎优化）

Optimize for Featured Snippets, People Also Ask, Knowledge Panels:

Agent prompt:

Optimize for Featured Snippets — structure content with clear question headers, concise answers (40-60 words), and bulleted/numbered lists.

Pattern:

markdown

undefined

针对特色片段、相关问题、知识面板进行优化：

Agent指令:

Optimize for Featured Snippets — structure content with clear question headers, concise answers (40-60 words), and bulleted/numbered lists.

优化模式:

markdown

undefined

What is [Topic]?

[Topic] is [concise 40-60 word definition].

Key Benefits:

Benefit one
Benefit two
Benefit three

undefined

Benefit one
Benefit two
Benefit three

undefined

10. SEO Links

10. SEO链接分析

Link profile analysis — internal links, backlinks, anchor text, orphan pages:

python

undefined

链接 profile 分析 — 内部链接、反向链接、锚文本、孤立页面：

python

undefined

Extract all links from page

python3 scripts/parse_html.py example_com.html --extract-links

Agent analyzes: internal link structure, anchor text distribution, orphan pages


**Agent prompt:**

Audit internal linking structure for https://example.com — identify orphan pages, check anchor text diversity, analyze link depth, and suggest hub pages.

undefined


**Agent指令:**

Audit internal linking structure for https://example.com — identify orphan pages, check anchor text diversity, analyze link depth, and suggest hub pages.

undefined

11. SEO Programmatic

11. SEO程序化页面审计

Quality gates for programmatic SEO pages:

Agent prompt:

Review programmatic SEO pages for quality — check for thin content, duplicate patterns, template issues, and unique value signals on each page.

Quality gates:

Minimum 800 words unique content per page
No placeholder text (e.g., "[City]", "Lorem ipsum")
Unique titles and meta descriptions
3+ internal links to related pages

程序化SEO页面的质量管控：

Agent指令:

Review programmatic SEO pages for quality — check for thin content, duplicate patterns, template issues, and unique value signals on each page.

质量管控标准:

每页至少800字原创内容
无占位文本（如"[City]", "Lorem ipsum"）
独特的标题和元描述
至少3个指向相关页面的内部链接

12. SEO Competitors

12. SEO竞争对手分析

Comparison and alternatives page generation:

Agent prompt:

Generate a "Competitors" or "Alternatives" comparison page for [Product] — include feature tables, pricing comparison, use case fit, and neutral tone.

生成竞品对比或替代方案页面：

Agent指令:

Generate a "Competitors" or "Alternatives" comparison page for [Product] — include feature tables, pricing comparison, use case fit, and neutral tone.

13. SEO Hreflang

13. SEO Hreflang分析

International SEO and hreflang validation:

python

undefined

国际SEO及hreflang标签验证：

python

undefined

Parse hreflang tags

python3 scripts/parse_html.py example_com.html --extract-hreflang


**Agent prompt:**

Validate hreflang implementation for multi-language site — check bidirectional links, correct language codes (ISO 639-1), regional variants (en-US, en-GB), and x-default fallback.

undefined

python3 scripts/parse_html.py example_com.html --extract-hreflang


**Agent指令:**

Validate hreflang implementation for multi-language site — check bidirectional links, correct language codes (ISO 639-1), regional variants (en-US, en-GB), and x-default fallback.

undefined

14. SEO Plan

14. SEO策略规划

Strategic SEO planning with topical clusters and industry templates:

Agent prompt:

Create an SEO content strategy for a SaaS product — build topical clusters, identify pillar pages, keyword themes, and internal linking structure.

Industry templates available:

SaaS
E-commerce
Local Business
Publisher/Media
Agency
Generic

基于主题集群和行业模板制定SEO战略：

Agent指令:

Create an SEO content strategy for a SaaS product — build topical clusters, identify pillar pages, keyword themes, and internal linking structure.

可用行业模板:

SaaS
电商
本地商家
出版商/媒体
代理机构
通用模板

15. SEO GitHub

15. SEO GitHub优化

GitHub repository SEO optimization:

python

undefined

GitHub仓库SEO优化：

python

undefined

Generate GitHub SEO report

python3 scripts/github_seo_report.py owner/repo

Outputs:

- GITHUB-SEO-REPORT.md

- GITHUB-ACTION-PLAN.md


**Agent prompt:**

Optimize GitHub repository SEO for [owner/repo] — improve description, add topics, enhance README structure, optimize title/headline, and benchmark against query results.


**Key checks:**
- Repository description (160 chars, keyword-rich)
- Topics (8-12 relevant tags)
- README structure (clear headline, features, installation)
- Social preview image
- Community health files (CODE_OF_CONDUCT.md, CONTRIBUTING.md)


**Agent指令:**

Optimize GitHub repository SEO for [owner/repo] — improve description, add topics, enhance README structure, optimize title/headline, and benchmark against query results.


**关键检查项:**
- 仓库描述（160字符，含关键词）
- 主题标签（8-12个相关标签）
- README结构（清晰标题、功能介绍、安装说明）
- 社交预览图
- 社区健康文件（CODE_OF_CONDUCT.md, CONTRIBUTING.md）

16. SEO Article

16. SEO文章优化

Article data extraction and LLM-driven content optimization:

Agent prompt:

Extract article content and optimize — check headline, readability, keyword density, internal links, meta description, and suggest improvements.

文章数据提取及LLM驱动的内容优化：

Agent指令:

Extract article content and optimize — check headline, readability, keyword density, internal links, meta description, and suggest improvements.

10 Specialist Agents

10个专业Agent

The skill includes specialized agents for domain-specific analysis:

Technical SEO Agent — Crawlability, indexability, security, mobile, JS rendering
Content Quality Agent — E-E-A-T scoring, AI content detection
Performance Agent — Core Web Vitals (LCP, INP, CLS)
Schema Markup Agent — JSON-LD validation and generation
Sitemap Agent — XML sitemap quality gates
Visual Analysis Agent — Screenshots, above-the-fold analysis (requires Playwright)
GitHub Analyst Agent — Repository metadata and README optimization
GitHub Benchmark Agent — Query ranking and competitor intelligence
GitHub Data Agent — API fallback and traffic archival
Verifier Agent (Global) — Deduplication and contradiction suppression

该Skill包含用于特定领域分析的专业Agent：

Technical SEO Agent — 可抓取性、可索引性、安全性、移动端适配、JS渲染分析
Content Quality Agent — E-E-A-T评分、AI内容检测
Performance Agent — Core Web Vitals（LCP、INP、CLS）分析
Schema Markup Agent — JSON-LD验证与生成
Sitemap Agent — XML站点地图质量管控
Visual Analysis Agent — 截图、首屏分析（需Playwright）
GitHub Analyst Agent — 仓库元数据与README优化
GitHub Benchmark Agent — 查询排名与竞品情报
GitHub Data Agent — API fallback与流量归档
Verifier Agent (Global) — 去重与矛盾抑制

Key Scripts Reference

核心脚本参考

Evidence Collection

证据收集

python

undefined

python

undefined

Fetch page with SEO crawler headers

python3 scripts/fetch_page.py https://example.com

Output: example_com.html

Parse HTML for all SEO signals

python3 scripts/parse_html.py example_com.html

Extracts: title, meta, headings, links, images, schema, canonical

Check robots.txt and crawler policies

python3 scripts/robots_checker.py https://example.com

Multi-page crawl

python3 scripts/crawl_audit.py https://example.com --max-depth 2 --max-pages 50

undefined

python3 scripts/crawl_audit.py https://example.com --max-depth 2 --max-pages 50

undefined

Performance & Core Web Vitals

性能与Core Web Vitals

python

undefined

python

undefined

PageSpeed Insights + Core Web Vitals

python3 scripts/pagespeed.py https://example.com

Returns: LCP, INP, CLS scores + recommendations

Note: Requires PAGESPEED_API_KEY environment variable

Get free key: https://developers.google.com/speed/docs/insights/v5/get-started

undefined

undefined

Indexability & Crawlability

可索引性与可抓取性

python

undefined

python

undefined

Comprehensive indexability matrix

python3 scripts/indexability_matrix.py https://example.com

Checks: robots.txt, meta robots, X-Robots-Tag, canonical, status codes, sitemaps

Sitemap validation

python3 scripts/sitemap_checker.py https://example.com

Validates: URL limits, 404s, lastmod, discovery

undefined

undefined

Schema Validation

Schema验证

python

undefined

python

undefined

Validate JSON-LD schema

python3 scripts/validate_schema.py page.html

Checks: syntax, required fields, deprecated types, placeholders

undefined

undefined

Report Generation

报告生成

python

undefined

python

undefined

Full audit with all outputs

python3 scripts/audit_runner.py https://example.com

Generate HTML dashboard

python3 scripts/generate_report.py audit_results.json --output seo-report.html

Verify findings (deduplicate + prioritize)

python3 scripts/finding_verifier.py audit_results.json

undefined

python3 scripts/finding_verifier.py audit_results.json

undefined

Configuration

配置

Environment Variables

环境变量

bash

undefined

bash

undefined

PageSpeed Insights API key (optional but recommended)

export PAGESPEED_API_KEY=your_api_key_here

GitHub token for repository analysis (optional)

export GITHUB_TOKEN=your_github_token_here

undefined

export GITHUB_TOKEN=your_github_token_here

undefined

Script Dependencies

脚本依赖

Install Python dependencies:

bash

undefined

安装Python依赖：

bash

undefined

From project root

pip install -r requirements.txt

Or with installer flag

bash install.sh --target claude --install-deps


**Core dependencies:**
- `requests` — HTTP requests
- `beautifulsoup4` — HTML parsing
- `lxml` — Fast XML/HTML parser
- `playwright` (optional) — Visual analysis and screenshots

Install Playwright browsers:

```bash
playwright install chromium

bash install.sh --target claude --install-deps


**核心依赖:**
- `requests` — HTTP请求
- `beautifulsoup4` — HTML解析
- `lxml` — 快速XML/HTML解析器
- `playwright`（可选） — 可视化分析与截图

安装Playwright浏览器：

```bash
playwright install chromium

Or use installer flag

bash install.sh --target claude --install-deps --install-playwright

undefined

bash install.sh --target claude --install-deps --install-playwright

undefined

Common Patterns

常见使用模式

Pattern 1: Quick Page Audit

模式1：快速页面审计

python

undefined

python

undefined

1. Fetch page

python3 scripts/fetch_page.py https://example.com/page

2. Parse HTML

python3 scripts/parse_html.py example_com_page.html > page_data.json

3. Check Core Web Vitals

python3 scripts/pagespeed.py https://example.com/page > cwv_data.json

4. Agent analyzes JSON outputs with LLM reasoning


**Agent prompt:**

Analyze page_data.json and cwv_data.json — identify Critical/Warning/Pass findings, apply confidence labels, prioritize by impact, output structured action plan following llm-audit-rubric.md.

undefined


**Agent指令:**

Analyze page_data.json and cwv_data.json — identify Critical/Warning/Pass findings, apply confidence labels, prioritize by impact, output structured action plan following llm-audit-rubric.md.

undefined

Pattern 2: Multi-Page Crawl Audit

模式2：多页面抓取审计

python

undefined

python

undefined

1. Crawl site (respects robots.txt)

python3 scripts/crawl_audit.py https://example.com --max-depth 3 --max-pages 100 > crawl_results.json

2. Generate indexability matrix

python3 scripts/indexability_matrix.py https://example.com > indexability.json

3. Agent identifies issues: orphan pages, redirect chains, indexability blocks

undefined

undefined

Pattern 3: Schema Validation & Generation

模式3：Schema验证与生成

python

undefined

python

undefined

1. Parse existing schema

python3 scripts/parse_html.py page.html --extract-schema > existing_schema.json

2. Validate

python3 scripts/validate_schema.py page.html

3. Agent generates improved schema based on page content and validation errors

undefined

undefined

Pattern 4: GitHub Repository SEO

模式4：GitHub仓库SEO优化

python

undefined

python

undefined

Generate full GitHub SEO report

python3 scripts/github_seo_report.py owner/repo

Outputs:

- GITHUB-SEO-REPORT.md (current state + benchmarks)

- GITHUB-ACTION-PLAN.md (prioritized improvements)


**Agent workflow:**
1. Check repository metadata (description, topics, social preview)
2. Analyze README structure and keyword optimization
3. Benchmark title against search results
4. Suggest query-aligned title improvements
5. Identify missing community health files


**Agent工作流程:**
1. 检查仓库元数据（描述、主题标签、社交预览图）
2. 分析README结构与关键词优化情况
3. 对比搜索结果基准测试标题
4. 提出与查询匹配的标题优化建议
5. 识别缺失的社区健康文件

Pattern 5: Full Site Audit with Reports

模式5：全站审计与报告生成

bash

undefined

bash

undefined

One command generates all outputs

python3 scripts/audit_runner.py https://example.com

Generates:

1. audit_results.json (structured data)

2. audit_report.html (interactive dashboard)

3. FULL-AUDIT-REPORT.md (markdown report)

4. ACTION-PLAN.md (prioritized tasks)

undefined

undefined

LLM Audit Rubric

LLM审计准则

All SEO analyses must follow the standard rubric at

resources/references/llm-audit-rubric.md

:

所有SEO分析必须遵循

resources/references/llm-audit-rubric.md

中的标准准则：

Finding Format

发现格式

markdown

undefined

markdown

undefined

[Severity]: [Finding Title]

Evidence: [Specific, verifiable proof — URLs, code snippets, metrics]

Impact: [SEO consequence — traffic, rankings, crawlability, user experience]

Confidence:

Confirmed / Likely / Hypothesis

Fix: [Actionable, specific remediation steps]

Priority:

Impact: High / Medium / Low
Effort: High / Medium / Low

undefined

Evidence: [Specific, verifiable proof — URLs, code snippets, metrics]

Impact: [SEO consequence — traffic, rankings, crawlability, user experience]

Confidence:

Confirmed / Likely / Hypothesis

Fix: [Actionable, specific remediation steps]

Priority:

Impact: High / Medium / Low
Effort: High / Medium / Low

undefined

Severity Levels

严重级别

Critical — Blocks indexing, causes crawl errors, security issues
Warning — Degrades SEO performance, suboptimal implementation
Pass — Meets best practices
Info — Opportunities, neutral observations

Critical — 阻碍索引、导致抓取错误、安全问题
Warning — 降低SEO性能、实现方式不够优化
Pass — 符合最佳实践
Info — 优化机会、中性观察

Confidence Labels

置信度标签

Confirmed — Direct evidence (status codes, parsed HTML, API responses)
Likely — Strong inference from multiple signals
Hypothesis — Reasoned speculation requiring validation

Confirmed — 直接证据（状态码、解析后的HTML、API响应）
Likely — 多个信号推导的强结论
Hypothesis — 合理推测，需验证

Troubleshooting

故障排除

Script Errors

脚本错误

Issue:

ModuleNotFoundError: No module named 'requests'

Fix:

bash

pip install -r requirements.txt

问题:

ModuleNotFoundError: No module named 'requests'

解决方法:

bash

pip install -r requirements.txt

Or

pip install requests beautifulsoup4 lxml


---

**Issue:** `playwright._impl._errors.Error: Executable doesn't exist`

**Fix:**
```bash
pip install playwright
playwright install chromium

Issue: PageSpeed API rate limit (403 or 429 errors)

Fix:

bash

undefined

pip install requests beautifulsoup4 lxml


---

**问题:** `playwright._impl._errors.Error: Executable doesn't exist`

**解决方法:**
```bash
pip install playwright
playwright install chromium

问题: PageSpeed API请求受限（403或429错误）

解决方法:

bash

undefined

Get free API key (100 queries/day)

获取免费API密钥（每日100次查询）

https://developers.google.com/speed/docs/insights/v5/get-started

export PAGESPEED_API_KEY=your_key_here python3 scripts/pagespeed.py https://example.com

undefined

export PAGESPEED_API_KEY=your_key_here python3 scripts/pagespeed.py https://example.com

undefined

Installation Issues

安装问题

Issue: Skill not detected after installation

Fix:

bash

undefined

问题: 安装后Skill未被识别

解决方法:

bash

undefined

Verify installation path

验证安装路径

ls -la ~/.claude/skills/seo/

Restart IDE

重启IDE

Claude Code: restart extension

Claude Code: 重启扩展

Cursor: restart application

Cursor: 重启应用

Re-run installer with verbose output

重新运行安装脚本并开启详细输出

bash install.sh --target claude -v


---

**Issue:** Windows PowerShell execution policy error

**Fix:**
```powershell

bash install.sh --target claude -v


---

**问题:** Windows PowerShell执行策略错误

**解决方法:**
```powershell

Run once per session

每个会话运行一次

Set-ExecutionPolicy -Scope Process -ExecutionPolicy Bypass

Then run installer

然后运行安装脚本

.\install.ps1 --online --target claude

undefined

.\install.ps1 --online --target claude

undefined

Audit Quality Issues

审计质量问题

Issue: Too many low-confidence findings

Fix:

python

undefined

问题: 低置信度发现过多

解决方法:

python

undefined

Run finding verifier to deduplicate and prioritize

运行发现验证工具去重并排序

python3 scripts/finding_verifier.py audit_results.json

Agent: focus on Confirmed/Likely findings first, mark Hypothesis clearly

Agent: 优先关注Confirmed/Likely发现，明确标记Hypothesis


---

**Issue:** Missing evidence in audit reports

**Fix:**
- Always collect evidence first (`fetch_page.py`, `parse_html.py`, `pagespeed.py`)
- Reference specific URLs, line numbers, metric values
- Use `read_url_content` MCP tool before reasoning
- Apply llm-audit-rubric.md standard for every finding


---

**问题:** 审计报告中缺少证据

**解决方法:**
- 始终先收集证据（`fetch_page.py`, `parse_html.py`, `pagespeed.py`）
- 引用具体URL、行号、指标数值
- 推理前使用`read_url_content` MCP工具
- 每个发现都遵循llm-audit-rubric.md标准

Reference Data

参考数据

The skill includes up-to-date reference files (checked by CI):

Core Web Vitals — LCP (2.5s), INP (200ms), CLS (0.1)
E-E-A-T Framework — September 2025 Quality Rater Guidelines
Schema.org Types — Active, restricted, deprecated lists
Content Quality Gates — Word count minimums, readability thresholds
Google SEO Quick Reference — Best practices snapshot
LLM Audit Rubric — Standardized output contract

Reference freshness is validated with:

bash

python3 scripts/reference_freshness.py resources/references --max-age-days 90

该Skill包含实时更新的参考文件（由CI验证）：

Core Web Vitals — LCP（2.5秒）、INP（200毫秒）、CLS（0.1）
E-E-A-T框架 — 2025年9月质量评估指南
Schema.org类型 — 可用、受限、已弃用列表
内容质量标准 — 最低字数、可读性阈值
Google SEO快速参考 — 最佳实践摘要
LLM审计准则 — 标准化输出规范

参考文件新鲜度验证：

bash

python3 scripts/reference_freshness.py resources/references --max-age-days 90

Advanced Usage

高级用法

Custom Industry Strategy

自定义行业策略

python

undefined

python

undefined

Agent generates topical cluster strategy

Agent生成主题集群策略

Example for SaaS product:

SaaS产品示例:

Pillar page: /seo-tools

支柱页面: /seo-tools

Cluster pages:

集群页面:

/seo-tools/keyword-research

/seo-tools/link-building

/seo-tools/technical-audit

Each cluster page links back to pillar

每个集群页面链接回支柱页面

Agent uses industry template from:

Agent使用以下路径中的行业模板:

resources/skills/seo-plan.md

undefined

undefined

Multi-Language Hreflang

多语言Hreflang标签

html

<!-- Agent generates hreflang tags -->
<link rel="alternate" hreflang="en" href="https://example.com/page" />
<link rel="alternate" hreflang="es" href="https://example.com/es/page" />
<link rel="alternate" hreflang="de" href="https://example.com/de/page" />
<link rel="alternate" hreflang="x-default" href="https://example.com/page" />

Validation checklist:

Bidirectional links (each page references all others)
Correct ISO 639-1 language codes
Regional variants (en-US, en-GB) where applicable
x-default fallback for unmatched locales
Self-referential link included

html

<!-- Agent生成hreflang标签 -->
<link rel="alternate" hreflang="en" href="https://example.com/page" />
<link rel="alternate" hreflang="es" href="https://example.com/es/page" />
<link rel="alternate" hreflang="de" href="https://example.com/de/page" />
<link rel="alternate" hreflang="x-default" href="https://example.com/page" />

验证清单:

双向链接（每个页面引用所有其他语言版本）
正确的ISO 639-1语言代码
适用地区变体（en-US、en-GB）
x-default作为未匹配地区的 fallback
包含自引用链接

Programmatic SEO Quality Gates

程序化SEO质量管控

python

undefined

python

undefined

Agent validates programmatic pages before deployment

Agent在部署前验证程序化页面

quality_gates = { "min_words": 800, "unique_title": True, "unique_meta_description": True, "no_placeholders": True, "min_internal_links": 3, "unique_content_ratio": 0.7 # 70% unique vs template }

quality_gates = { "min_words": 800, "unique_title": True, "unique_meta_description": True, "no_placeholders": True, "min_internal_links": 3, "unique_content_ratio": 0.7 # 70%原创内容 vs 模板内容 }

Reject pages that fail gates

拒绝未通过管控标准的页面

Log failures for content team review

记录失败情况供内容团队审核

undefined

undefined

Wiki & Documentation

维基与文档

Full documentation available at: https://github.com/Bhanunamikaze/Agentic-SEO-Skill/wiki

Key wiki pages:

Script Inventory — All 89 scripts with purpose notes
Installation Guide — Detailed per-IDE setup
Example Prompts — 50+ tested agent prompts
Troubleshooting — Common issues and fixes
Report Generation — Custom report templates

完整文档请访问: https://github.com/Bhanunamikaze/Agentic-SEO-Skill/wiki

关键维基页面:

脚本清单 — 全部89个脚本及用途说明
安装指南 — 针对不同IDE的详细设置步骤
示例指令 — 50+经过测试的Agent指令
故障排除 — 常见问题及解决方法
报告生成 — 自定义报告模板

License

许可证

MIT License — free for commercial and personal use.

MIT许可证 — 可免费用于商业和个人用途。

agentic-seo-skill

Original

Translation

Agentic SEO Skill

Agentic SEO Skill

Installation

安装

Quick Install (No Cloning Required)

快速安装（无需克隆）

Install to all supported IDEs

Install to all supported IDEs

Install to Claude Code only

Install to Claude Code only

Install to specific project

Install to specific project

From Source

从源码安装

Install to Claude Code with dependencies

Install to Claude Code with dependencies

Install with Playwright for visual analysis

Install with Playwright for visual analysis

Installation Targets

安装目标

Core Workflow

核心工作流程

16 Specialized Sub-Skills

16项专业子技能

1. SEO Audit (Full Site)

1. SEO审计（全站）

Quick full audit with all outputs

Quick full audit with all outputs

Outputs:

Outputs:

- audit_results.json

- audit_results.json

- audit_report.html

- audit_report.html

- FULL-AUDIT-REPORT.md

- FULL-AUDIT-REPORT.md

- ACTION-PLAN.md

- ACTION-PLAN.md

2. SEO Page (Single Page Analysis)

2. SEO页面分析（单页）

Fetch page content

Fetch page content

Parse HTML for SEO signals

Parse HTML for SEO signals

Check Core Web Vitals

Check Core Web Vitals

3. SEO Technical

3. SEO技术审计

Check robots.txt policies

Check robots.txt policies

Run indexability matrix (robots, meta robots, canonicals, status codes)

Run indexability matrix (robots, meta robots, canonicals, status codes)

Multi-page crawl audit

Multi-page crawl audit

4. SEO Content

4. SEO内容审计

Extract article content

Extract article content

Run content quality checks (agent analyzes with LLM)

Run content quality checks (agent analyzes with LLM)

Check: word count, readability, E-E-A-T signals, topical depth

Check: word count, readability, E-E-A-T signals, topical depth

5. SEO Schema

5. SEO Schema分析

Validate existing schema markup

Validate existing schema markup

Check for deprecated schema types

Check for deprecated schema types

Agent generates schema based on page content

Agent generates schema based on page content

6. SEO Sitemap

6. SEO站点地图分析

Discover and validate sitemaps

Discover and validate sitemaps

Check for issues: 404s, lastmod quality, URL limits (50,000 per file)

Check for issues: 404s, lastmod quality, URL limits (50,000 per file)

7. SEO Images