How to add an llms.txt file to your site

What an llms.txt file actually is

It is a proposed convention, not a rule. You put a Markdown file at yoursite.com/llms.txt and fill it with links to the pages you most want a model to understand: your docs, your pricing, whatever explains what you do. Each link gets a one-line description. That is the whole idea. Give a model a clean map so it does not have to reverse-engineer your site from a cluttered navigation bar.

Jeremy Howard of Answer.AI proposed the format in 2024. You can read the full spec on the llms.txt glossary page, but the shape is simple enough to hold in your head, which is part of the appeal.

What goes inside the file

The format is ordinary Markdown. Start with an H1 that names your site or product. Follow it with a short blockquote that says, in one or two sentences, what you are. Then group your links under H2 headings, and give every link a one-line note about what the page is for. Here is a small, complete example:

# Acme Analytics

> Acme is a privacy-first web analytics tool for small SaaS teams.

## Docs
- [Getting started](https://acme.com/docs/start): install the snippet and see your first numbers
- [API reference](https://acme.com/docs/api): endpoints, auth, and rate limits

## Product
- [Pricing](https://acme.com/pricing): plans and what each tier includes
- [Security](https://acme.com/security): how we handle data and where it lives

## Optional
- [llms-full.txt](https://acme.com/llms-full.txt): the same pages, expanded into one file

That last entry points at an optional companion file. If you want to hand over more than a list of links, you can add an llms-full.txt that holds fuller content, say your docs stitched into one long page. Most sites do not need it. A tidy llms.txt is plenty to start.

How to add it, step by step

• Write the file. Open a text editor, follow the shape above, and save it as llms.txt. Keep it to the pages that genuinely matter. A short, honest list beats a dump of every URL you own.
• Put it at the root. The file lives at yoursite.com/llms.txt, the same place your robots.txt sits. On most hosts that means dropping it in your public or static folder. On Next.js, a file in /public/llms.txt is served at the root automatically.
• Check it loads. Visit the URL in a browser. You should see raw text, not a 404 and not your styled homepage. If a framework rewrites the path, add an exception so the file is served as-is.
• Keep the links honest as the site changes. A page you moved six months ago should not still be in the file. Stale links are worse than no file, since they point a reader at something that no longer exists.

Will anything actually read it?

This is the part people skip, so read it twice. As of today, no major AI company has publicly committed to reading llms.txt. Whether a crawler like GPTBot fetches the file and does something useful with it is up to the company that runs it, and none of them have promised they will. So set your expectations at the floor. Adding one might help a little. It might do nothing measurable.

That is not a reason to skip it. It is a reason to treat it as what it is: low-cost housekeeping. Ten minutes, no downside, a small chance of upside if and when these tools start honoring the convention. Just do not put it on the list of things you expect to change your numbers, and do not let a vendor sell it to you as one.

What it does not do

An llms.txt file does not make your site crawlable. It does not rewrite thin pages into good ones. It does not clean up messy HTML that a model struggles to parse. It only makes the pages you already have easier to find, and it can only point at work that is already worth pointing at.

A fair way to think about it

Add the file. It costs you an afternoon at most and probably ten minutes. Then go spend your real effort on the things that move how often AI names you: clear answers to the questions your buyers ask, a consistent story across the sites a model already reads, and evidence a model can check. If you want to see where you stand before you start, a free AI visibility audit shows you what the major engines say about you today, which is a better use of the next hour than polishing a file nobody has promised to read.