metehan.ai Research & SOPs

Appearance

Screaming Frog AI Scripts ​

Fresh

This content reflects Metehan Alp's Screaming Frog custom scripts as of 2025-2026.

Six custom JavaScript scripts for Screaming Frog SEO Spider that integrate AI analysis directly into technical SEO crawls. Each script extends Screaming Frog's functionality with AI-powered insights.

Prerequisites ​

  • Screaming Frog SEO Spider (latest version)
  • Configuration > Custom > JavaScript enabled
  • API keys for respective AI services (where applicable)

Scripts ​

1. Visual Query Fan-Out ​

Generates graphical visualizations of how AI search engines decompose queries into sub-queries. Produces relationship diagrams showing query hierarchies and overlap patterns.

What it does:

  • Crawls target pages and extracts topic signals
  • Maps predicted sub-query decompositions
  • Generates visual graphs showing sub-query relationships
  • Exports relationship data as structured JSON

GitHub: github.com/nicche

Use with: Query Fan-Out Methodology

2. Query Fan-Out Analysis ​

The data extraction companion to Visual Query Fan-Out. Produces raw position matrices and RRF scores instead of visualizations.

What it does:

  • Extracts SERP position data for sub-query clusters
  • Builds position matrices (URL x sub-query)
  • Calculates RRF fused scores per URL
  • Identifies coverage gaps where your domain has no presence
  • Flags pages that meet or miss the citation threshold (tau = 0.020)

GitHub: github.com/nicche/query-fan-out-analysis

Use with: RRF Top-n Playbook

3. MUVERA Analysis ​

MUVERA (Multi-Vector Retrieval Analysis) examines how content is represented across multiple vector dimensions in retrieval systems. This script analyzes content for multi-vector alignment with AI retrieval models.

What it does:

  • Analyzes page content across multiple embedding dimensions
  • Scores content alignment with predicted retrieval vectors
  • Identifies sections with weak vector representation
  • Recommends content adjustments for improved retrieval

GitHub: github.com/nicche

Implementation notes:

  • Requires API access to an embedding service (e.g., OpenAI, Gemini)
  • Processing time scales with page content length
  • Best used on pillar content and high-value pages

4. DeepSeek Integration ​

Integrates DeepSeek AI models into Screaming Frog crawls for content analysis and optimization recommendations.

What it does:

  • Sends crawled page content to DeepSeek API
  • Returns AI-generated content analysis per page
  • Scores content quality, topic coverage, and E-E-A-T signals
  • Generates optimization recommendations during the crawl
  • Outputs results as custom extraction columns

GitHub: github.com/nicche

Implementation notes:

  • Requires DeepSeek API key
  • Rate limiting applies (configure delay between requests)
  • Cost scales with number of pages crawled and content length

5. LLMO Optimization Analyzer ​

Scores pages for Large Language Model Optimization (LLMO) readiness. Analyzes structural, semantic, and authority signals that influence LLM citation probability.

What it does:

  • Scores heading structure for LLM readability
  • Evaluates content chunk sizes against optimal token ranges
  • Checks for answer-first content structure
  • Analyzes entity coverage and topical completeness
  • Scores schema markup presence and quality
  • Generates an overall LLMO readiness score (0-100) per page

GitHub: github.com/nicche

Implementation notes:

  • No external API required (rule-based analysis)
  • Fast execution, suitable for large crawls
  • Export LLMO scores as a custom column for sorting and filtering

6. Automated Backlink Outreach ​

Uses AI to generate personalized outreach emails from backlink and link prospect data discovered during crawls.

What it does:

  • Identifies broken links, resource pages, and link opportunities during crawl
  • Extracts contact information where available
  • Generates personalized outreach email drafts using AI
  • Matches your content to the link opportunity context
  • Exports email drafts with prospect data in a ready-to-send format

GitHub: github.com/nicche

Implementation notes:

  • Requires AI API key for email generation (GPT-4 or similar)
  • Review all generated emails before sending (AI drafts need human review)
  • Works best when combined with a link prospecting crawl configuration

Installation ​

For all scripts:

  1. Download the script from the GitHub repository
  2. Open Screaming Frog > Configuration > Custom > JavaScript
  3. Click "Add" and paste the script content
  4. Configure API keys and parameters in the script header
  5. Save and run your crawl

Screaming Frog Version

These scripts are tested with Screaming Frog v20+. Earlier versions may not support all JavaScript features used.

All Repositories ​

All scripts are maintained under Metehan's GitHub organization: github.com/nicche