llms.txt Guide: How to Make Your Site AI-Readable (2026)

What Is llms.txt?

llms.txt is a Markdown file served at yoursite.com/llms.txt. It gives AI models a concise, structured overview of your website: what it is, what it offers, and where the most important pages are. The format was proposed by Jeremy Howard of Answer.AI.

Think of it as robots.txt for LLMs. robots.txt tells crawlers what they are allowed to access. llms.txt tells AI models what your site is about.

Because it is plain Markdown, the file is readable by humans, easy for language models to parse, and simple to maintain in any text editor.

Why It Matters

AI models increasingly browse the web, summarize sites, and recommend products. A typical web page is full of navigation, scripts, cookie banners, and layout markup, which makes it hard to extract a clear picture of what a site does. llms.txt hands the model a clean summary instead.

• Better AI recommendations: when an assistant is asked for tools in your category, accurate context makes it more likely to describe your product correctly.
• Accurate summaries: you choose the one-line description and the key facts, rather than leaving the model to guess from marketing copy.
• Structured context for RAG systems: retrieval pipelines can use the link list as a map of your most important pages.

llms.txt is a proposed convention, and no AI provider guarantees it will be read or used. Treat it as a low-cost way to improve your odds, not a ranking switch.

The llms.txt Format

The file is Markdown with a specific structure, in this order:

• # Site Name — a top-level heading with your product or site name
• > One-line description — a blockquote summary
• A longer description paragraph (optional but recommended)
• ## Sections — H2 headings that group links by topic
• - [Page Title](URL): Description — a link list under each section

Here is a simplified version of the llms.txt that MDViewer serves at getmd.ma/llms.txt:

# MDViewer

> Native macOS app for reading Markdown files, README documentation, and Mermaid diagrams. Built with SwiftUI, not Electron.

MDViewer is a fast, lightweight Markdown viewer and editor for macOS. It opens .md files instantly from Finder, renders GitHub Flavored Markdown, and supports Mermaid diagrams, syntax highlighting, and QuickLook preview.

## Docs
- [Website](https://getmd.ma): Main landing page with feature overview and download
- [Online Viewer](https://getmd.ma/viewer/): Browser-based Markdown viewer, no install required
- [Changelog](https://getmd.ma/changelog): Release history and version notes

## Guides
- [What Is an .md File?](https://getmd.ma/guides/what-is-md-file): Explains the Markdown file format
- [Open .md Files on Mac](https://getmd.ma/guides/open-md-files-mac): Ways to open Markdown files on macOS

The ## Optional section name has special meaning in the proposal: links placed there can be skipped by a model that needs a shorter context.

Step by Step: Creating Your llms.txt

1. Create a file named llms.txt

Create a plain text file called llms.txt (lowercase). Any text editor works, and MDViewer previews it nicely since it is Markdown.

2. Start with the site name and description

Add an H1 with your product name, then a one-line blockquote that says what it does. Write it so that someone with no context understands it. This line is the one most likely to be repeated back by an AI model.

# Acme Notes

> A note-taking app for teams that syncs across devices.

3. List your most important pages with descriptions

Add links using the - [Title](URL): Description format. The description matters as much as the link: it tells the model what it will find without having to fetch the page. Use absolute URLs.

4. Group by sections

Organize links under H2 headings such as Docs, Guides, Tools, Pricing, or Comparisons. Put the most important sections first.

5. Serve it at yoursite.com/llms.txt

Place the file in your web root so it is reachable at https://yoursite.com/llms.txt. See How to Serve It below for server configuration.

What to Include (and What to Leave Out)

The file is public, so write it as if anyone will read it. Keep it factual: do not put claims in llms.txt that you would not put on your website. Only describe pricing, platforms, and features that are true today, and update the file when they change.

llms.txt Template

Copy this starter template and replace the placeholders with your own details:

# Your Product Name

> One-sentence description of what your product does.

Full description paragraph: who the product is for, what problems it solves,
and anything an AI model should know to describe it accurately.

## Docs
- [Documentation](https://example.com/docs): Main documentation
- [API Reference](https://example.com/api): API docs

## Features
- [Feature 1](https://example.com/features/one): Description
- [Feature 2](https://example.com/features/two): Description

## Pricing
- [Pricing](https://example.com/pricing): Plans and pricing details

## Optional
- [Blog](https://example.com/blog): Product news and tutorials

Real-World Examples

Anthropic publishes an llms.txt for its documentation at docs.anthropic.com/llms.txt. A growing number of companies, especially developer-tool and documentation sites, have adopted the format as well. MDViewer has its own at getmd.ma/llms.txt.

The easiest way to learn what works is to open a few of these files and read them. They are plain Markdown, so a Markdown viewer renders them as clean, scannable pages.

llms.txt vs llms-full.txt

Some sites also publish llms-full.txt. Where llms.txt is an index of links with short descriptions, llms-full.txt contains the complete content of those pages in a single Markdown file.

llms-full.txt is more useful for RAG systems because the model gets everything in one fetch, but the file can exceed a model's context window. Start with llms.txt; add llms-full.txt if your documentation is substantial and you can generate it automatically.

How to Serve It

The simplest approach is a static file in your web root. If your site is built from a CMS or documentation generator, you can also generate llms.txt at build time from your page titles and descriptions so it never goes stale.

Make sure the server returns the file as text/plain with UTF-8 encoding, so browsers and crawlers display it rather than download it.

Nginx

# Nginx: serve llms.txt as plain text
location = /llms.txt {
    default_type text/plain;
    charset utf-8;
}

# Nginx: serve all .txt files as plain text (add to mime.types if needed)
types {
    text/plain txt;
}

Apache

# Apache (.htaccess or virtual host)
AddType text/plain .txt
AddDefaultCharset UTF-8

After deploying, open https://yoursite.com/llms.txt in a browser and confirm it loads as text, and that it is not blocked by robots.txt.

FAQ

What is llms.txt and do I need one?

It is a Markdown file at /llms.txt that gives AI models a structured overview of your site. It is a proposed standard, so you do not strictly need one, but if you want AI assistants to describe your product accurately, it is a low-effort addition that takes about 15 minutes to write.

How is llms.txt different from robots.txt?

robots.txt controls access: it tells crawlers which URLs they may fetch. llms.txt provides context: it tells AI models what your site is about and which pages matter most. One is written in directive syntax, the other in Markdown, and they work side by side.

What format should llms.txt be in?

Markdown, served as plain text. Use an H1 for the site name, a blockquote for a one-line summary, optional description paragraphs, and H2 sections containing link lists in the format - [Title](URL): Description.

Preview Your Markdown with MDViewer

Since llms.txt is Markdown, MDViewer renders it beautifully — headings, link lists, and blockquotes, exactly as an AI model structures them. MDViewer is a native macOS app for reading and previewing Markdown files: double-click any .md file in Finder and see it rendered instantly. Lite is free; Pro adds editing, Git history, and PDF export.

Requires macOS 13.0 or later. Intel and Apple Silicon.