llms.txt

ADP provides machine-readable documentation following the llms.txt standard. This allows AI assistants (Claude, ChatGPT, Cursor, Copilot, and others) to quickly understand the project and provide accurate answers about it.

What is llms.txt?

llms.txt is an open standard proposed by Jeremy Howard (fast.ai). It provides a well-known URL (/llms.txt) where websites expose their documentation in a clean markdown format optimized for LLM consumption — no HTML, no JavaScript, no navigation clutter.

Available Files

File	URL	Content
llms.txt	/llms.txt	Concise table of contents with links to per-page .md files
llms-full.txt	/llms-full.txt	All documentation pages concatenated into a single file
*.md (per-page)	e.g. /guide/collectors.md	Clean markdown for any individual page

llms.txt is a lightweight index — titles and links. Use it when your AI tool has limited context or you need a quick overview.

llms-full.txt contains the complete documentation in one file (~13K tokens). Use it when your AI tool has a large context window and you want comprehensive answers.

Per-page .md files are available for every documentation page. Use them when you need information about a specific topic without loading the entire documentation.

How to Use

In Claude (claude.ai)

Paste the URL in the chat:

Read https://app-dev-panel.github.io/app-dev-panel/llms-full.txt and answer my questions about ADP.

In Claude Code

Use the WebFetch tool or paste the URL:

Fetch https://app-dev-panel.github.io/app-dev-panel/llms-full.txt

In Cursor / Copilot / other IDE assistants

Add the URL as a documentation source in your IDE settings, or paste it into the chat context.

In custom agents

Fetch the file programmatically and include it in the system prompt:

import httpx

response = httpx.get("https://app-dev-panel.github.io/app-dev-panel/llms-full.txt")
docs = response.text

messages = [
    {"role": "system", "content": f"ADP documentation:\n\n{docs}"},
    {"role": "user", "content": "How do I add a custom collector?"},
]

What's Inside

Both files are auto-generated at build time from the same markdown sources that produce this documentation site. They include:

• Getting Started — installation for each framework
• Architecture — layers, data flow, proxy system
• Collectors — all 30 collectors with descriptions
• API Reference — REST endpoints, SSE, Inspector API
• Adapters — Symfony, Yii 2, Yii 3, Laravel
• MCP Server — AI integration setup
• CLI — commands and options

Content stays in sync automatically — when a documentation page is updated, the next build regenerates both files.

How It Works

Powered by vitepress-plugin-llms — the same plugin used by Vite, Vue.js, and Vitest.

The plugin operates as a Vite plugin during the build:

1. Collects all markdown files during Vite's transform phase
2. Cleans content via remark AST — strips frontmatter, HTML, Vue components
3. Generates llms.txt (TOC), llms-full.txt (concatenated), and per-page .md files
4. Injects hidden hints on HTML pages pointing LLMs to the .md versions

Only English pages are included. Russian translations are excluded via ignoreFiles: ['ru/**'].

Source: plugin config in config.ts