How to Perform a Deep Single-Page GEO Analysis with Claude Code

Use the /geo page <url> slash command to execute a focused, single-page Generative Engine Optimization audit that outputs a composite score and detailed markdown report.

Conducting a deep single-page GEO analysis allows you to validate individual URLs—such as product landing pages or campaign articles—before rolling out changes across your entire site. The zubair-trabzada/geo-seo-claude repository provides a specialized command that orchestrates multiple AI agents to evaluate technical signals, content quality, and AI-search readiness for any specific URL.

The /geo page Command Entry Point

The analysis pipeline is triggered through a dedicated slash command defined in [geo/SKILL.md](https://github.com/zubair-trabzada/geo-seo-claude/blob/main/geo/SKILL.md).

/geo page <url>

When invoked, this command initiates a six-stage pipeline that mimics a full-site audit but restricts all evaluations to the supplied URL only. Unlike traditional SEO crawlers, this workflow specifically targets Generative Engine Optimization signals—factors that influence how AI systems like ChatGPT, Perplexity, and Google AI Overviews cite and represent your content.

The Six-Stage Analysis Pipeline

Stage 1: Page Retrieval via fetch_page.py

The foundation of the audit is [scripts/fetch_page.py](https://github.com/zubair-trabzada/geo-seo-claude/blob/main/scripts/fetch_page.py), a helper script that retrieves the raw HTML without executing JavaScript. It extracts the initial page payload, including the HTML structure and HTTP headers, which feeds directly into the evaluation agents.

Stage 2: Content and Signal Extraction

From the raw HTML, the pipeline derives critical on-page elements:

• Page title and meta description
• Heading hierarchy (H1-H6 structure)
• Word count and content density
• JSON-LD schema blocks (detected via the geo-schema sub-skill)
• AI-crawler accessibility signals (robots.txt status, llms.txt presence)

Stage 3: Parallel Sub-Agent Evaluation

The command invokes five specialized sub-agents simultaneously, each defined in the agents/ directory. Every agent receives only the single page’s extracted data:

• AI Visibility ([agents/geo-ai-visibility.md](https://github.com/zubair-trabzada/geo-seo-claude/blob/main/agents/geo-ai-visibility.md)): Calculates citability scoring and checks for brand mention opportunities.
• Platform Optimization ([agents/geo-platform-analysis.md](https://github.com/zubair-trabzada/geo-seo-claude/blob/main/agents/geo-platform-analysis.md)): Assesses readiness for ChatGPT, Perplexity, Google AI Overviews, and other generative platforms.
• Technical GEO ([agents/geo-technical.md](https://github.com/zubair-trabzada/geo-seo-claude/blob/main/agents/geo-technical.md)): Verifies SSR (Server-Side Rendering) presence, Core Web Vitals potential, and security headers.
• Content E-E-A-T ([agents/geo-content.md](https://github.com/zubair-trabzada/geo-seo-claude/blob/main/agents/geo-content.md)): Evaluates Experience, Expertise, Authority, and Trustworthiness signals.
• Schema & Structured Data ([agents/geo-schema.md](https://github.com/zubair-trabzada/geo-seo-claude/blob/main/agents/geo-schema.md)): Validates application/ld+json blocks for semantic accuracy.

These agents share the scoring methodology documented in the repository’s scoring-methodology document, ensuring consistent evaluation criteria.

Stage 4: Composite GEO Score Calculation

After individual evaluations, the system aggregates scores using weighted categories:

• Citability: 25%
• Brand: 20%
• E-E-A-T: 20%
• Technical GEO: 15%
• Schema: 10%
• Platform Optimization: 10%

The resulting composite score ranges from 0 to 100 and represents the page’s overall AI-search readiness.

Stage 5: Report Generation

The pipeline writes GEO-PAGE-ANALYSIS.md to the current working directory. This file follows the same structure as full-site audit reports—containing an executive summary, score breakdown, deep-dive sections, and prioritized quick-wins—but remains scoped to the single inspected URL.

For PDF export, the repository uses the HTML template located at [templates/geo-report-template.html](https://github.com/zubair-trabzada/geo-seo-claude/blob/main/templates/geo-report-template.html) to render the markdown into a styled document.

Stage 6: Usage Feedback

Upon completion, Claude Code prints a concise inline summary displaying the overall GEO score, key strengths, and the highest-impact quick win. This allows immediate action without requiring you to open the markdown file first.

Command Line Examples

Execute a basic single-page analysis:

claude-code /geo page https://example.com/product/awesome-gadget

Pipe the output for terminal viewing:

claude-code /geo page https://example.com/blog/ai-search-trends | less

Generate a PDF version using Pandoc:

claude-code /geo page https://example.com/blog/ai-search-trends \
  && pandoc GEO-PAGE-ANALYSIS.md -o GEO-PAGE-ANALYSIS.pdf

The PDF conversion leverages the same Pandoc-based workflow employed by the geo-report-pdf skill, utilizing the HTML template for consistent styling.

Summary

• The /geo page <url> command initiates a focused audit defined in geo/SKILL.md.
• The scripts/fetch_page.py script retrieves raw HTML without JavaScript execution to extract structured data.
• Five parallel sub-agents evaluate AI Visibility, Platform Optimization, Technical GEO, Content E-E-A-T, and Schema signals.
• A weighted composite score (0-100) combines Citability (25%), Brand (20%), and E-E-A-T (20%) metrics.
• Output includes both an inline summary and a detailed GEO-PAGE-ANALYSIS.md report.
• The templates/geo-report-template.html file provides the styling scaffold for PDF exports.

Frequently Asked Questions

What is the difference between a single-page GEO analysis and a full-site audit?

A single-page GEO analysis utilizes the same five specialized sub-agents and scoring methodology as a full-site audit, but restricts all data inputs to one specific URL. This is ideal for validating new landing pages or campaign content before broader rollout, whereas full-site audits crawl multiple URLs to identify systemic issues across your domain.

Does the fetch_page.py script execute JavaScript when crawling?

No. According to the source code in scripts/fetch_page.py, the script retrieves raw HTML only without JavaScript execution. This means dynamically rendered content (such as JavaScript-heavy SPAs) will be evaluated based on their server-side rendered markup or initial HTML payload.

How is the final GEO score calculated?

The composite score is calculated using weighted percentages from five evaluation categories: Citability (25%), Brand presence (20%), E-E-A-T signals (20%), Technical GEO factors (15%), Schema markup (10%), and Platform Optimization (10%). Each category is scored independently by specialized agents in the agents/ directory before being aggregated into the 0-100 scale.

Can I export the single-page analysis as a PDF?

Yes. After generating GEO-PAGE-ANALYSIS.md, you can convert it to PDF using Pandoc with the repository’s HTML template. The command pandoc GEO-PAGE-ANALYSIS.md -o GEO-PAGE-ANALYSIS.pdf utilizes templates/geo-report-template.html for styling, ensuring the PDF matches the visual format of full-site audit reports.

{
  "mcpServers": {
    "instagit": {
      "command": "npx",
      "args": ["-y", "instagit@latest"]
    }
  }
}