llms.txt

Overview

HolyDocs automatically generates llms.txt and llms-full.txt files during each build. These files follow the llms.txt specification and provide a structured, machine-readable version of your documentation optimized for large language model consumption.

What is llms.txt?

The llms.txt standard provides a way for websites to offer their content in a format optimized for LLM consumption. Instead of requiring AI models to parse HTML, the .txt files provide clean, structured text with navigation metadata.

Generated Files

HolyDocs generates four AI-native files during each build:

The generated /skill.md file describes your specific documentation site for agent discovery. It is different from the public HolyDocs agent skills repository at seangeng/holydocs-skills, which you install into coding assistants with npx skills add.

llms.txt Format

The llms.txt file contains a summary of your documentation structure:

text
# My API Docs> Documentation for the My API service.## Getting Started- [Introduction](/introduction): Overview of the API and key concepts.- [Quickstart](/quickstart): Get your first API call working in 5 minutes.- [Authentication](/api/authentication): API keys, OAuth, and JWT authentication.## API Reference- [Projects](/api/projects): Create and manage projects.- [Projects](/api/projects): Create and manage projects.

llms-full.txt Format

The llms-full.txt file includes the complete text content of every page:

text
# My API Docs> Documentation for the My API service.## IntroductionOverview of the API and key concepts.The My API service provides a RESTful API for managing...[Full page content continues...]---## QuickstartGet your first API call working in 5 minutes.[Full page content continues...]

Accessing the Files

The files are served by the renderer at your documentation domain:

text
https://docs.yourcompany.com/llms.txthttps://docs.yourcompany.com/llms-full.txthttps://docs.yourcompany.com/skill.mdhttps://docs.yourcompany.com/.well-known/agent-card.json

Contextual Menu

HolyDocs includes a contextual menu button (bottom-left of every page) that provides quick access to AI-native features:

Additional options include opening in Cursor, copying the MCP server URL, and linking to the llms.txt file.

Agent Card (ERC-8004)

The agent card at /.well-known/agent-card.json follows the ERC-8004 specification for AI agent discovery:

json
{  "name": "My API Docs",  "description": "Documentation for the My API service",  "url": "https://docs.yourcompany.com",  "capabilities": {    "mcp": "https://api.holydocs.com/api/v1/mcp/PROJECT_ID/message",    "search": "https://api.holydocs.com/api/v1/docs/PROJECT_ID/search",    "llmsTxt": "https://docs.yourcompany.com/llms.txt"  }}

The llms.txt files are regenerated on every production deployment. There is no separate configuration needed — if your docs build successfully, the AI-native files are automatically created and served.

Storage

All AI-native files are stored in R2 alongside your built documentation. They are served with appropriate content types:

• .txt files: text/plain; charset=utf-8
• .md files: text/markdown; charset=utf-8
• .json files: application/json