LLMS.txt Generator - Airbais AI Tools

Overview

LLMS.txt is a proposed standard that is intended to aid in context engineering for website content. The LLMS.txt Generator crawls websites and creates standardized LLMS.txt files following the specification at llmstxt.org.

Key Features

Universal Compatibility

Automatically detects website structure and content categories for any type of site

Dynamic Section Detection

Intelligently discovers content sections from URL patterns rather than using hardcoded templates

Multiple Output Formats

Generates .txt, .md, and .json versions with comprehensive metadata

Dashboard Integration

Automatically integrates with the master dashboard for visual analysis and reporting

Getting Started

Clone the repository

git clone https://github.com/Airbais/airbais-tools.git
cd tools/llmstxtgenerator

Install dependencies

pip install -r requirements.txt

Set up API keys (optional)

For AI-enhanced descriptions:

export OPENAI_API_KEY="your-openai-api-key"
# OR
export ANTHROPIC_API_KEY="your-anthropic-api-key"

Run your first generation

python llmstxtgenerator.py https://example.com --dashboard

Quick Examples

Basic Usage
With Custom Details
Using Configuration
View Results

# Generate LLMS.txt for a website
python llmstxtgenerator.py https://example.com

Configuration

Customize the tool’s behavior through config.yaml:

Website Settings

Configure target website and basic information:

website:
  url: "https://example.com"
  name: null  # Auto-detected if not specified
  description: null  # Auto-detected or AI-generated

Generation Settings

Control crawling behavior and section detection:

generation:
  max_pages: 100  # Maximum pages to crawl
  max_depth: 3  # Maximum crawl depth
  min_pages_per_section: 2  # Minimum pages to create a section
  max_links_per_section: 20  # Maximum links per section
  ignore_segments:  # URL segments to ignore for section detection
    - "p"
    - "c"
    - "s"
    - "id"
    - "category"
    - "page"

Content Analysis

Configure AI-powered content analysis:

analysis:
  use_ai_descriptions: false  # Generate descriptions with AI
  ai_model: "gpt-4o-mini"  # AI model to use
  extract_meta: true  # Extract meta tags
  extract_headings: true  # Extract page headings

Output Options

Control output formats and organization:

output:
  directory: "results"
  keep_past_results: true
  generate_report: true
  formats:  # Output formats to generate
    - "txt"
    - "markdown"
    - "json"

How It Works

Website Discovery

Crawls the target website respecting robots.txt and following links systematically

Dynamic Section Detection

Analyzes URL patterns to automatically discover content categories and sections

Content Extraction

Extracts titles, descriptions, and metadata from each discovered page

Intelligent Categorization

Groups pages by detected sections with configurable filtering and limits

LLMS.txt Generation

Creates structured output following the official LLMS.txt specification

Multi-Format Export

Generates .txt, .md, and .json versions with comprehensive metadata

Dashboard Integration

Creates dashboard-compatible data for visual analysis and reporting

Dynamic Section Detection

Unlike template-based approaches, the tool discovers sections directly from your website’s URL structure

URL Pattern Analysis

Analyzes URL paths to identify natural content groupings and hierarchies

Smart Filtering

Ignores common URL segments like IDs, pagination, and utility pages

Universal Compatibility

Works with any website type: documentation, e-commerce, blogs, or corporate sites

Output Structure

Results are organized by date for easy historical tracking:

results/
└── YYYY-MM-DD/
    ├── llms.txt              # Standard LLMS.txt file
    ├── llms.md               # Markdown version with metadata
    ├── llms.json             # Structured JSON data
    ├── generation_report.md  # Detailed generation report
    └── dashboard-data.json   # Dashboard integration data

Example Output

# FashionShop: Clothes, Shoes & Accessories For Women, Men & Kids

> Shop DressMeUp.com for the Highest Quality Women's and Men's Clothing and see the entire selection of Children's Clothing, Cashmere Sweaters, Women's Dresses and Shoes, Men's Suits, Jackets, Accessories and more.

## Company
- [Military & Medical Discount](https://fashionshop.com/company/military-medical-first-responder-discount): Discounts for service members
- [FashionShop Credit Card](https://fashionshop.com/company/credit-card): Store credit card information
- [Mobile App](https://fashionshop.com/company/the-fashionshop-mobile-app): Download the FashionShop app

## Help
- [Contact Us](https://fashionshop.com/help/contact-us): Customer service and support
- [Shipping & Handling](https://fashionshop.com/help/shipping-handling): Delivery information
- [Order Status](https://fashionshop.com/help/order-status): Track your orders

## Shop
- [Women's Sale](https://fashionshop.com/shop/cashmere-summer-sale/womens): End of season cashmere sale
- [Men's Sale](https://fashionshop.com/shop/cashmere-summer-sale/mens): Men's cashmere sale items
- [Top Rated](https://fashionshop.com/shop/top-rated/mens): Customer favorite products

Dashboard Features

Two dashboard options: Local tool-specific dashboard and Master multi-tool dashboard

Visual Analytics

Generation Summary

Success rates, pages crawled, and sections discovered with key metrics

Site Structure Analysis

Interactive visualization of discovered content categories and sections

Content Distribution

Breakdown of pages by section with detailed statistics

Configuration Overview

Current settings and parameters used for generation

Local Dashboard
Master Dashboard

# Launch after generation
python llmstxtgenerator.py https://example.com --dashboard

# View existing results
python llmstxtgenerator.py --dashboard-only

Tool-specific: Shows only LLMS.txt generator results
Fast Loading: Optimized for single-tool analysis
Detailed Metrics: Complete generation statistics and analysis

Command Line Reference

url

string

Website URL to generate LLMS.txt for (optional if using config file)

--config

string

Configuration file path (default: config.yaml)

--name

string

Website/project name (overrides auto-detection)

--description

string

Website/project description (overrides auto-detection)

--max-pages

number

Maximum number of pages to crawl

--max-depth

number

Maximum crawl depth

--no-ai

flag

Disable AI-generated descriptions

--output-dir

string

Output directory for results

--dashboard

flag

Launch dashboard after generation

--dashboard-only

flag

Launch dashboard without running generation

--verbose

flag

Enable verbose logging

--version

flag

Show program version number

Real-World Examples

E-commerce Site
Documentation Site
Corporate Website

# FashionShop example - discovers shopping categories
python llmstxtgenerator.py https://fashionshop.com

Result: Sections like shop, sale, help, company automatically detected from URL patterns like /shop/, /sale/, /help/, /company/Success Rate: 47% with 7 sections discovered from 100 pages

Performance Guidelines

Processing time scales with website size and complexity

Small Sites (<50 pages)

Processing Time: 1-3 minutes
Recommended Settings: Default configuration works well
Success Rate: Typically 60-80%

Medium Sites (50-200 pages)

Processing Time: 3-10 minutes
Optimization: Consider increasing min_pages_per_section to 3-5
Success Rate: Typically 40-60%

Large Sites (200+ pages)

Processing Time: 10+ minutes
Optimization: Set max_pages: 100-200 to limit scope
Success Rate: Typically 30-50% (more selective)

Troubleshooting

Low Success Rate

Symptoms: Less than 20% success rate or very few sections detectedSolutions:

Lower min_pages_per_section in config
Check if the site has a consistent URL structure
Verify the site allows crawling (check robots.txt)
Try increasing max_depth for deeper crawling

No Pages Found

Symptoms: “No pages crawled” or empty resultsSolutions:

Verify URL is accessible and correct
Check internet connection
Ensure site doesn’t block automated crawlers
Try adjusting user_agent in crawling configuration

AI Descriptions Not Working

Symptoms: Generic descriptions despite enabling AISolutions:

Verify API key is set: echo $OPENAI_API_KEY
Check API key has sufficient credits/quota
Try different AI model in configuration
Use --no-ai flag to disable and test basic functionality

Dashboard Not Loading

Symptoms: Dashboard shows no data or fails to startSolutions:

Ensure generation completed successfully
Check dashboard-data.json exists in results directory
Verify no JSON formatting errors in results
Try local dashboard: python llmstxtgenerator.py --dashboard-only

Advanced Usage

Custom URL Patterns

Control which URLs to include or exclude:

generation:
  include_patterns:
    - "https://example.com/docs/.*"
    - "https://example.com/api/.*"
  exclude_patterns:
    - ".*\\.(pdf|zip|tar\\.gz)$"
    - ".*/downloads/.*"
    - ".*/temp/.*"

Section Detection Tuning

Fine-tune automatic section detection:

generation:
  min_pages_per_section: 3  # Require more pages per section
  ignore_segments:          # Additional segments to ignore
    - "temp"
    - "staging"
    - "preview"
  max_links_per_section: 15  # Limit section size

Without AI Enhancement

For faster generation without AI descriptions:

python llmstxtgenerator.py https://example.com --no-ai

Requirements

System Requirements

Python 3.8+
256MB RAM minimum
Internet connection for crawling
Optional: API keys for AI-enhanced descriptions

Dependencies

Key Python packages (automatically installed):

requests - Web crawling and HTTP requests
beautifulsoup4 - HTML parsing and content extraction
pyyaml - Configuration file handling
openai - AI-powered descriptions (optional)
anthropic - Alternative AI provider (optional)
plotly - Dashboard visualization
pandas - Data processing and analysis

AI Tools Suite Integration

LLMS.txt Generator is part of the larger Airbais AI Tools Suite with centralized dashboard

Master Dashboard

Centralized view of all AI tool results at ../dashboard/

Auto-Discovery

New tools are automatically detected and integrated

Standard Format

JSON output compatible with other suite tools

Consistent Design

Shared Airbais design system across all tools

Suite Benefits

Unified Analysis: Combine LLMS.txt generation with intent analysis and brand evaluation
Historical Tracking: Compare website changes over time across all tools
Streamlined Workflow: Single dashboard for all AI-powered website analysis
Future-Ready: Architecture designed for easy addition of new analysis tools

Contributing

We welcome contributions in these key areas:

Section Detection

Improved algorithms for automatic content categorization

Content Extraction

Better handling of complex website structures and content types

Output Formats

Additional export formats and integration capabilities

Performance

Optimization for large-scale websites and faster processing

Development Guidelines

Maintain Universal Compatibility: Ensure changes work across different website types
Follow Dynamic Detection: Enhance the URL pattern-based section discovery
Preserve Dashboard Integration: Keep dashboard-data.json format compatible
Add Comprehensive Testing: Test with real websites of different types
Update Documentation: Keep this MDX file current with new features

Future Roadmap

Enhanced AI Integration

Better content summarization and automatic description generation

Real-time Updates

Monitor websites for changes and automatically update LLMS.txt files

API Integration

Direct integration with popular CMS platforms and website builders

Multi-language Support

Generate LLMS.txt files for international websites with language detection

Advanced Analytics

Content quality scoring and SEO optimization recommendations

Part of the Airbais AI Tools Suite - Comprehensive tools for AI-powered business intelligence and content optimization.

Get Started

What's New

Tools

Automation

Learn

​Overview

​Key Features

Universal Compatibility

Dynamic Section Detection

Multiple Output Formats

Dashboard Integration

​Getting Started

​Quick Examples

​Configuration

​How It Works

​Dynamic Section Detection

URL Pattern Analysis

Smart Filtering

Universal Compatibility

​Output Structure

​Example Output

​Dashboard Features

​Visual Analytics

Generation Summary

Site Structure Analysis

Content Distribution

Configuration Overview

​Command Line Reference

​Real-World Examples

​Performance Guidelines

​Troubleshooting

​Advanced Usage

​Custom URL Patterns

​Section Detection Tuning

​Without AI Enhancement

​Requirements

System Requirements

​Dependencies

​AI Tools Suite Integration

Master Dashboard

Auto-Discovery

Standard Format

Consistent Design

​Suite Benefits

​Contributing

Section Detection

Content Extraction

Output Formats

Performance

​Development Guidelines

​Future Roadmap

Overview

Key Features

Getting Started

Quick Examples

Configuration

How It Works

Dynamic Section Detection

Output Structure

Example Output

Dashboard Features

Visual Analytics

Command Line Reference

Real-World Examples

Performance Guidelines

Troubleshooting

Advanced Usage

Custom URL Patterns

Section Detection Tuning

Without AI Enhancement

Requirements

Dependencies

AI Tools Suite Integration

Suite Benefits

Contributing

Development Guidelines

Future Roadmap