Audit

GEO Readiness Audit

Performs a complete GEO (Generative Engine Optimization) readiness audit on any URL. The agent executes each check using its own tools — no external CLI required.

How to Run

# Option A: Use the CLI
python3 cli/geo-max audit <url>

# Option B: Run the script directly
python3 scripts/audit.py <url>

# Option C: Let the agent do it manually using the instructions below

Agent Instructions (do this manually if no CLI)

When the user asks "audit this URL" and the scripts aren't installed, perform these checks using your own tools:

Check 1: llms.txt (20 pts)

1. Fetch https://<domain>/.well-known/llms.txt and https://<domain>/llms.txt using URL fetch
2. If either exists, score 20 pts. Note the content quality: does it list important URLs with descriptions?
3. Also check https://<domain>/robots.txt — does it allow GPTBot, ClaudeBot, Google-Extended?

Check 2: Schema.org Structured Data (25 pts)

1. Fetch the URL and extract all application/ld+json script blocks
2. Validate that @context is https://schema.org and @type is a valid type
3. Score breakdown:
   • 25 pts: 3+ valid schema types present (Article, FAQPage, Organization, Product, etc.)
   • 15 pts: 1-2 valid types
   • 5 pts: schema exists but invalid
   • 0 pts: no schema

Check 3: EEAT Signals (25 pts)

1. Check HTML meta tags: <meta name="author">
2. Check for byline pattern: by FirstName LastName in text
3. Check for publish date: datePublished, dateModified, or time element
4. Check for external reference links (links to other domains)
5. Check for social/credential links (LinkedIn, Twitter, GitHub)
6. Score: 25 for strong author + date + references, 15 for partial, 5 for weak

Check 4: AI Crawler Access (15 pts)

1. Fetch robots.txt
2. Check if GPTBot, ClaudeBot, Google-Extended, PerplexityBot are explicitly allowed
3. Check for any blanket Disallow: / that blocks crawlers
4. Score: 15 if all allowed, 10 if some, 5 if restricted, 0 if blocked

Check 5: Content Structure (15 pts)

1. Check for clear H1/H2 heading hierarchy
2. Check first 500 chars for statistics, data, or concrete claims (cites sources)
3. Check word count (1000+ words preferred for AI extraction)
4. Score based on depth

Final Score

• 90-100: Excellent — strong AI search presence
• 70-89: Good — minor gaps
• 50-69: Needs work — several missing signals
• 0-49: Poor — invisible in AI search

Output Format

Return results as a formatted markdown report:

## GEO Audit Report for https://example.com

**Score: 68/100 — Needs work**

### 1. llms.txt — 0/20 ❌
No llms.txt found at /.well-known/llms.txt or /llms.txt. Create one listing your key pages.

### 2. Schema.org — 20/25 ✅
Found: Article, Organization, FAQPage (3 types, all valid)
Recommendation: Add Product or HowTo schema for relevant pages.

### 3. EEAT Signals — 15/25 ⚠️
- Author meta tag found: ✓
- Publish date: ✓
- External references: 2 (min 3 recommended)
- Author credentials/social: missing LinkedIn

### 4. AI Crawler Access — 10/15 ⚠️
- GPTBot: ✓ allowed
- ClaudeBot: ✓ allowed
- Google-Extended: not mentioned
- PerplexityBot: not mentioned

### 5. Content — 8/15 ⚠️
~800 words, 1 statistic in first 500 chars. Add more data-backed claims.

Related Checks

• EEAT deep dive: geo-max audit-eeat <url>
• Schema validation: geo-max audit-schema <url>
• Full report: Save to geo-audit-report.json