How to write an llms.txt file for your website (the new GEO standard)

llms.txt is the emerging convention for telling AI assistants what's on your site, which pages matter, and how to cite them. Here's the 10-minute setup.

6 min read intermediateUpdated Jun 23, 2026

By Billionideas Editorial Team

Reviewed by the editorial team · Jun 23, 2026

Inspired by robots.txt, llms.txt is a single Markdown file at the root of your domain that hand-curates the URLs AI assistants should read first. Anthropic, Mintlify and Vercel have already adopted it.

It's not yet officially recognised by Google or OpenAI, but ChatGPT, Claude and several agentic crawlers do honour it — and the cost of adding one is essentially zero.

Before you start

The ability to upload a file to the root of your domain (public folder for most static sites)

Step by step

01
Create the file
Create llms.txt in your project's public/ or static/ folder. It must be served at https://yourdomain.com/llms.txt.
02
Add the H1 with your project name
First line: '# [Your project name]'. Second line: a one-sentence summary blockquote. This is what AI engines quote when introducing your site.
03
Add a short Description section
An H2 'Description' followed by 2–3 sentences explaining what the site does, who it's for and what's on it.
04
List your key pages under H2 sections
Group URLs by section — 'Docs', 'Guides', 'API'. Each link as a Markdown bullet: '- [Title](https://...) : one-line summary'.
05
Mark optional sections clearly
An H2 'Optional' section for nice-to-have pages (blog posts, changelogs). LLMs treat these as lower-priority context.
06
Add the file to your sitemap
Reference llms.txt in your sitemap.xml so crawlers discover it. Some agentic tools also check robots.txt for an llms.txt directive.
07
Optional: add llms-full.txt
A longer, fuller version with the actual content (not just links). Great for documentation sites — Mintlify generates this automatically.