Composer Installation 

composer require web-vision/ai-llms-txt

Basic TypoScript Setup 

The extension automatically includes its TypoScript configuration. No manual setup is required for basic functionality.

Site Configuration 

The extension automatically uses your site's configuration from config/sites/[site]/config.yaml. If you want to customize the routing, you can add custom route enhancers:

imports:
  -
    resource: 'EXT:ai_llms_txt/Configuration/Routes/RouterEnhancer.yaml'

This import adds the following route enhancers:

• llms.txt for the llms.txt specification (typeNum 1699)
• .md suffix for Markdown content (typeNum 1701)

llms.txt Files 

• https://yoursite.com/?type=1699 - Default access via typeNum
• https://yoursite.com/llms.txt - Alternative direct access (if route enhancer is configured)

Markdown Content 

• https://yoursite.com/any-page.md - Markdown version of any TYPO3 page

llms.txt Files 

The extension automatically generates an llms.txt file that includes:

• Site title and description - Taken from your site's root page
• Navigation structure - Your page hierarchy and menu structure
• Topics and keywords - Configured by administrators
• Contact information - Configured by administrators

This file helps AI systems understand your site's structure and content.

Markdown Pages 

Any page on your website can be viewed in Markdown format by adding .md to the URL. For example:

• /about becomes /about.md
• /services/consulting becomes /services/consulting.md

The Markdown version includes:

• Page title - From the page properties
• Page description - From the page properties
• All content elements - Headers, text, images, etc. converted to Markdown format

Page Properties 

Use descriptive titles

Clear, descriptive page titles help AI systems understand your content's purpose.

Add page descriptions

The description field in page properties provides context for AI systems and appears in both llms.txt and Markdown output.

Structure your navigation logically

Your site's navigation structure is included in the llms.txt file, so logical organization helps AI systems understand your content hierarchy.

Content Elements 

Use semantic headers

Use header elements (H1, H2, H3, etc.) to structure your content logically.

Write clear, descriptive text

Well-written content is more useful for AI systems and human readers alike.

Add alt text to images

Image alternative text is included in Markdown conversion and helps AI systems understand visual content.

Organize content logically

Content elements are rendered in their page order, so logical organization improves the Markdown output.

llms.txt Page Type 

llmstxt = PAGE
llmstxt {
    typeNum = 1699

    config {
        disableAllHeaderCode = 1
        additionalHeaders.10.header = Content-Type: text/plain; charset=utf-8
        additionalHeaders.10.replace = 1
        xhtml_cleaning = 0
        admPanel = 0
        debug = 0
        no_cache = 1
    }

    10 = USER
    10 {
        userFunc = WebVision\AiLlmsTxt\Controller\LlmsTxtController->generateAction
    }
}

Markdown Page Type 

markdown_page = PAGE
markdown_page {
    typeNum = 1701

    config {
        disableAllHeaderCode = 1
        additionalHeaders.10.header = Content-Type: text/plain; charset=utf-8
        additionalHeaders.10.replace = 1
        xhtml_cleaning = 0
        admPanel = 0
        debug = 0
        no_cache = 1
        forceAbsoluteUrls = 1
    }

    10 = USER
    10 {
        userFunc = WebVision\AiLlmsTxt\Controller\LlmsTxtController->renderPageAsMarkdown
    }
}

Custom Navigation Filtering 

If you need to exclude certain pages from the llms.txt navigation structure, you can extend the NavigationBuilder service or use standard TYPO3 page properties:

• Set "Hide in navigation" to exclude pages from llms.txt
• Use "Access" settings to control visibility
• Set pages to hidden to exclude them entirely
• Set no index meta tag on pages to prevent inclusion in llms.txt

Custom Content Processing 

The extension uses TYPO3's standard content rendering. To customize how content appears in Markdown:

• Use standard TYPO3 content element configuration
• Customize TypoScript rendering for specific content types
• The extension respects all standard TYPO3 content visibility settings

Services 

ConfigurationService

Handles all site configuration reading and provides typed access to configuration values. Uses the TYPO3 core request injection pattern - request is set once via setRequest() and all methods access it internally.

LlmsTxtGeneratorService

Main application service that orchestrates the generation of llms.txt content by coordinating other services.

MarkdownConverterService

Converts HTML content to Markdown format using the league/html-to-markdown library.

NavigationBuilder

Builds the site navigation structure for inclusion in llms.txt files.

Repository 

PageRepository

Provides database access for page records with proper respect for TYPO3's language and workspace handling.

Controller 

LlmsTxtController

Thin controller layer that handles HTTP requests and delegates business logic to services. Provides entry points for TypoScript USER objects.

Controller Methods 

Service Classes 

Unit Testing 

The extension includes comprehensive unit tests, particularly for the ConfigurationService. Tests can be run against all supported TYPO3 versions:

# Run unit tests with TYPO3 v12
Build/Scripts/runTests.sh -s unit -t 12

# Run unit tests with TYPO3 v13
Build/Scripts/runTests.sh -s unit -t 13

# Run unit tests with TYPO3 v14
Build/Scripts/runTests.sh -s unit -t 14

# Run with specific PHP version
Build/Scripts/runTests.sh -s unit -t 14 -p 8.3

The ConfigurationService tests cover:

• Request injection and retrieval
• Fallback to $GLOBALS['TYPO3_REQUEST']
• All configuration getters (isEnabled, getMaxDepth, getTitleOverride, etc.)
• Default values and type casting

Functional Testing 

Functional tests can be run using:

Build/Scripts/runTests.sh -s functional

Markdown Conversion 

Complex HTML Structures

Some complex HTML structures may not convert perfectly to Markdown. This particularly affects:

• Multi-column layouts
• Nested tables with complex formatting
• Custom HTML elements with specific styling
• Interactive elements (JavaScript widgets, forms)

Content Element Limitations

Certain TYPO3 content elements may not convert optimally:

• File collections with custom rendering
• Media galleries with specific layouts
• Custom content elements without semantic HTML

Large Page Content

Pages with very large amounts of content may experience:

• Memory limitations during HTML-to-Markdown conversion
• Slower response times for .md requests
• Potential timeouts on resource-constrained servers

Navigation Structure 

Deep Navigation Hierarchies

Sites with very deep page hierarchies (6+ levels) may experience:

• Performance impacts during navigation building
• Large llms.txt files that may be difficult for AI systems to process
• Memory usage scaling with site complexity

Hidden Page Handling

The extension respects TYPO3's page visibility settings, but:

• Pages hidden in navigation are excluded from llms.txt
• Access-restricted pages may not appear in navigation structure
• Workspace/versioning states may affect page inclusion

Web Server Configuration 

.well-known Directory Access

Some web server configurations may block access to .well-known directories:

• Apache servers may require specific .htaccess rules
• Nginx servers may need location block configuration
• Some shared hosting providers block hidden directory access

MIME Type Handling

Text/plain MIME type for .md and llms.txt files may not be properly configured on all servers.

HTML-to-Markdown Edge Cases 

Issue: Nested blockquotes may not render correctly

Workaround: Avoid deeply nested blockquote structures in content elements

Issue: Table formatting may be simplified or lost

Workaround: Use simple table structures for content that will be converted to Markdown

Issue: Custom CSS classes and styling are not preserved

Expected Behavior: Markdown is a semantic format - visual styling is intentionally simplified

Performance Issues 

Issue: Large sites (1000+ pages) may experience slow llms.txt generation

Workaround: Reduce the maxDepth setting in TypoScript configuration

Solution: Consider implementing page-level caching for navigation structures

Issue: Memory usage scales with page content size

Workaround: Monitor memory limits and consider splitting very large pages

Solution: Implement chunked processing for extremely large content

Compatibility Issues 

Issue: Some third-party extensions may interfere with content rendering

Symptoms: Missing content in Markdown output or errors during generation

Workaround: Test with third-party extensions disabled to isolate conflicts

Issue: Custom TypoScript configurations may affect page type rendering

Symptoms: Incorrect MIME types or additional headers in output

Solution: Ensure the extension's TypoScript is loaded after custom configurations

Performance Enhancements 

• Caching Layer: Implementation of dedicated caching for navigation structures and frequently accessed content
• Chunked Processing: Support for processing very large pages in chunks to reduce memory usage
• Selective Rendering: Options to exclude specific content types from Markdown conversion

Feature Additions 

• Custom Content Filters: Configuration options to exclude specific content element types
• Enhanced Metadata: Support for additional llms.txt specification fields as the standard evolves
• Multi-language Support: Better handling of multi-language sites and language-specific llms.txt files

Developer Experience 

• PSR-14 Events: Addition of events for custom processing hooks
• Better Error Handling: More detailed error messages and logging
• Development Tools: CLI commands for testing and debugging llms.txt generation

Large Site Performance 

For sites with performance issues:

llmstxt.settings {
    # Reduce navigation depth
    maxDepth = 2

    # Consider adding custom page exclusion logic
    # (requires custom extension development)
}

Custom Content Filtering 

To exclude specific content types from Markdown conversion, extend the controller:

<?php
// Custom implementation to filter content elements
// See Developer documentation for detailed examples

Changes 

Added

• Unit tests for ConfigurationService with full coverage (24 tests)
• TYPO3 v14 LTS support in test runner scripts
• Comprehensive test suite runnable across TYPO3 12, 13, and 14

Changed

• Refactored ConfigurationService to use TYPO3 core request injection pattern
• Request is now injected once via setRequest() instead of passing to each method
• Added fallback to $GLOBALS['TYPO3_REQUEST'] for backward compatibility
• Improved code architecture following TYPO3 ContentObjectRenderer pattern

Removed

• Removed DownloadMarkdownCommand from Services.yaml (command not yet implemented)

Latest Changes 

Fixed * Corrected the exclusion of specific doktypes in PageRepository to ensure proper fetching of child pages.

Latest Changes 

Changed

• Update PageRepository to exclude specific doktypes and fetch their children

Release Date: 2025-10-29

Initial Release 

This is the initial release of the LLMS TXT Generator extension for TYPO3 v13.

composer require web-vision/ai-llms-txt