llms.txt is the file everyone in SEO suddenly has an opinion about. Here is the calm version: what it actually is, the exact format, whether anyone important reads it, and how to write one in five minutes — followed by an honest take on whether you should bother. Want to know if your site already has one? Run the checker below.
An llms.txt file is a single Markdown document that lives at the root of your domain, at yoursite.com/llms.txt. Its job is to point large language models at the pages on your site that are worth reading, in plain language they can parse easily. Where a sitemap lists every URL for a search crawler, llms.txt does the opposite — it is a short, curated shortlist meant for AI tools, with a little context attached to each link.
The format is deliberately simple. You open with an H1 that names the site, you can add a one-line summary as a blockquote, and then you group your important links under section headings. Nothing about it is enforced, and there is no validator that blocks you — it is a convention, not a standard with teeth.
A minimal, well-formed llms.txt looks like this:
# Acme Analytics
> Acme is a privacy-first web analytics platform. These links cover the
> product, the API and the most-read guides.
## Docs
- [Getting started](https://acme.com/docs/start): install and first dashboard
- [API reference](https://acme.com/docs/api): endpoints and auth
## Guides
- [Migrating from Google Analytics](https://acme.com/guides/ga-migration)
- [Cookieless tracking explained](https://acme.com/guides/cookieless)
That is the whole idea: an H1 title, an optional summary, then sections of annotated links. Some sites also publish an llms-full.txt that inlines the actual page content, but the short curated version is the common case.
This is where the honesty has to come in. After roughly eighteen months of conversation, adoption is sitting at about one site in ten, and it is concentrated in SaaS, developer tools and documentation. More importantly, Google has stated plainly that it does not use llms.txt and is not planning to, and at least one study of AI-citation data found the file added no predictive signal. Some AI tools and smaller agents do read it, so it is not pointless — but it is not the load-bearing wall some posts make it out to be.
For a documentation-heavy or content-heavy site, sure — it costs a few minutes, it cannot hurt, and it nudges the tools that do read it toward your best pages. Just keep the order of operations right. If an AI engine is blocked in your robots.txt, or your content only appears after JavaScript runs, or your pages carry no structured data, those problems will sink your AI visibility long before a missing llms.txt does. Fix the foundation first, then add llms.txt as a finishing touch.
The fastest way to see where you stand on all of that at once is the AI readiness checker: it reports whether your llms.txt exists and validates, whether ChatGPT, Perplexity, Google AI and Claude can crawl you, and whether your homepage is readable without JavaScript. If you only want to look at crawler access, the guide to controlling AI crawlers goes deeper on the robots.txt side.
llms.txt is a small Markdown file placed at the root of a site (yoursite.com/llms.txt) that curates the pages most worth reading for large language models. It starts with an H1 site name, an optional one-line summary, and then sections of links to your key documentation, articles or product pages. Think of it as a hand-picked table of contents written for AI tools rather than for search crawlers.
No. robots.txt controls access — which crawlers may fetch which URLs. llms.txt does not grant or deny anything; it is purely a suggestion of which existing pages matter most. The two sit side by side: robots.txt decides who gets in, llms.txt hints at what to read first.
No. Google has publicly said it does not use llms.txt and has no plans to. Some other AI tools and smaller agents do read it, but you should treat it as optional and supplementary, not as a ranking factor or a requirement.
It is cheap to add and harmless to have, so for documentation-heavy or content-heavy sites it can be a reasonable extra. Just keep the priorities straight: letting AI crawlers reach you, serving content without a JavaScript wall, and marking pages up with schema all matter far more for AI visibility than an llms.txt file does.