ai-readability-audit

AI Website Readability Audit

Audit your website from the perspective of a simulated AI crawler, checking whether structured data and Meta information are friendly to large models.

Why is an AI readability audit needed?

Traditional SEO audits focus on Googlebot, but in the AI search era, your website also needs to be friendly to LLMs:

ChatGPT Crawl web pages via Bing
Perplexity Directly crawl and cite websites
Claude Answer questions via search augmentation

If your website is not AI-friendly, you may lose traffic brought by AI search.

Trigger Conditions

Activate this skill when the user says the following:

"Check whether this website is AI-friendly"
"AI readability audit"
"LLM friendliness check"
"Website AI optimization check"

Workflow

┌─────────────────┐    ┌─────────────────┐    ┌─────────────────┐
│  输入网站URL     │───▶│  获取HTML内容    │───▶│  提取关键特征    │
│                 │    │  (WebFetch)     │    │  Meta/Schema等  │
└─────────────────┘    └─────────────────┘    └─────────────────┘
                                                      │
                                                      ▼
                              ┌─────────────────────────────────────┐
                              │           AI 可读性分析              │
                              │  • 结构化数据检查                    │
                              │  • 语义标签评估                      │
                              │  • 内容清晰度评分                    │
                              │  • 生成优化建议                      │
                              └─────────────────────────────────────┘

Execution Steps

Step 1: Get Web Page Content

Use WebFetch to get the complete HTML content of the target URL.

Step 2: Extract HTML Features

Analyze the following key elements:

Meta Information

<title>...</title>
<meta name="description" content="...">
<meta name="keywords" content="...">
<meta property="og:title" content="...">
<meta property="og:description" content="...">

Structured Data

<script type="application/ld+json">...</script>
<!-- Schema.org 标记 -->

Semantic Tags

<header>, <nav>, <main>, <article>, <section>, <aside>, <footer>
<h1>, <h2>, <h3>...

Content Structure

Whether the heading hierarchy is correct
Whether paragraph length is reasonable
Usage of lists and tables
Whether link text is descriptive

Step 3: AI Readability Assessment

Score according to the following dimensions (each item 0-10 points):

Dimension	Weight	Check Items
Structured Data	25%	Schema.org、JSON-LD、Open Graph
Semantic HTML	20%	Use of semantic tags, heading hierarchy
Meta Completeness	15%	title、description、keywords
Content Clarity	20%	Paragraph structure, location of key information
Technical Accessibility	20%	Readability without JS, robots.txt

Total Score Calculation: weighted average, full score 100 points

Step 4: Generate Audit Report

# AI 可读性审计报告

**网站**: [URL]
**审计时间**: YYYY-MM-DD HH:mm
**综合评分**: XX/100 ⭐⭐⭐☆☆

---

## 评分详情

| 维度 | 得分 | 状态 |
|------|------|------|
| 结构化数据 | X/10 | ✅/⚠️/❌ |
| 语义 HTML | X/10 | ✅/⚠️/❌ |
| Meta 完整性 | X/10 | ✅/⚠️/❌ |
| 内容清晰度 | X/10 | ✅/⚠️/❌ |
| 技术可访问性 | X/10 | ✅/⚠️/❌ |

---

## 检测结果

### ✅ 做得好的地方

1. [好的方面1]
2. [好的方面2]

### ❌ 需要改进的地方

1. **问题**: [问题描述]
   **影响**: [对AI可读性的影响]
   **建议**: [具体修复建议]

2. **问题**: [问题描述]
   **影响**: [对AI可读性的影响]
   **建议**: [具体修复建议]

---

## 详细分析

### 结构化数据

**检测到的 Schema 类型**:
- [x] Organization
- [ ] Article
- [ ] Product
- [ ] FAQ

**建议添加**:
```json
{
  "@context": "https://schema.org",
  "@type": "Article",
  "headline": "...",
  "author": {...}
}

Meta Tags

Tag	Current Value	Suggestion
title	[Current Value]	[Suggested Value]
description	[Current Value]	[Suggested Value]

Semantic Structure

Heading Hierarchy:

H1: [标题] ✅
├── H2: [子标题]
│   ├── H3: [小节]
│   └── H3: [小节]
└── H2: [子标题]

Issue: [if any issues]

Priority Fix List

Sorted by impact level:

🔴 High Priority: [Issue] - Estimated improvement X%
🟡 Medium Priority: [Issue] - Estimated improvement X%
🟢 Low Priority: [Issue] - Estimated improvement X%

AI Search Compatibility

AI Search Engine	Compatibility	Description
ChatGPT (Bing)	⭐⭐⭐	[Description]
Perplexity	⭐⭐⭐	[Description]
Google SGE	⭐⭐	[Description]

Report automatically generated by Claude Code


## 使用示例

### 示例 1：审计单个页面

User: Help me check https://example.com whether it is AI-friendly

Claude:

Get page HTML
Extract Meta, Schema, and other information
Evaluate scores across dimensions
Generate detailed audit report


### 示例 2：批量审计

User: Check the AI readability of these pages

Claude: [Audit each page in turn, generate a summary report]


## 数据存储

审计结果默认保存到：`~/.claude/cache/ai-readability-audit/`

建议保存以下文件（便于复查、对比与回归）：
- `audit-{域名}-{YYYYMMDD-HHMMSS}.md`：完整审计报告（可直接分享/粘贴）
- `audit-{域名}-{YYYYMMDD-HHMMSS}.json`：结构化结果（评分、问题清单、建议、关键证据）

可选（体积较大，按需开启）：
- `html-{域名}-{YYYYMMDD-HHMMSS}.html`：抓取到的原始 HTML（用于排查“抓取内容与页面不一致”）

## AI 可读性最佳实践

### 1. 必备的结构化数据

```html
<!-- 组织信息 -->
<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "Organization",
  "name": "公司名",
  "url": "https://example.com",
  "logo": "https://example.com/logo.png"
}
</script>

<!-- 文章页面 -->
<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "Article",
  "headline": "文章标题",
  "author": {"@type": "Person", "name": "作者名"},
  "datePublished": "2025-01-01"
}
</script>

2. Semantic HTML templates

<main>
  <article>
    <header>
      <h1>文章标题</h1>
      <p>文章摘要/描述</p>
    </header>

    <section>
      <h2>第一部分</h2>
      <p>内容...</p>
    </section>

    <footer>
      <p>作者信息、发布日期等</p>
    </footer>
  </article>
</main>

3. Put key information first

✅ 好的结构：
[核心结论/定义] - 第一段
[支持论据1]
[支持论据2]
[详细解释]

❌ 不好的结构：
[背景介绍]
[历史沿革]
[详细解释]
[最后才是结论]

Dependencies/Tools

WebFetch: Get webpage HTML content
Write: Save audit report

Limitations

Can only audit publicly accessible pages
SPA/JavaScript-rendered content may not be fully retrievable
Cannot detect restrictions on AI crawlers in robots.txt