What is llms.txt, and does it actually do anything?

llms.txt is a Markdown file at the root of a website (/llms.txt) that gives AI tools a title, a one-line summary and links to the site's most useful pages. It was proposed in September 2024 at llmstxt.org. So far, few AI systems request it: Ahrefs found that 97% of the llms.txt files it studied got no requests in May 2026. It's cheap to add, but it is not a way to rank in AI answers.

What is llms.txt for?

Web pages are built for people. They come with navigation, cookie banners, scripts and layout, and the useful part can be a small fraction of the HTML. A language model reading a site has a limited amount of text it can work with at once, so wading through all of that is wasteful.

The idea behind llms.txt is to hand AI tools a short, clean summary instead: what the site is, and where the pages that matter live, in plain Markdown. It sits alongside the files sites already publish for machines. robots.txt says what crawlers may fetch, the sitemap lists every page you want indexed, and llms.txt picks out the few pages an assistant or AI agent should read first.

What goes in an llms.txt file?

The proposal defines a simple structure, in this order:

1. A title as a level-1 Markdown heading, with the name of the site or project. This is the only part the proposal requires.
2. A short summary in a blockquote: one or two sentences on what the site is.
3. Optional paragraphs or lists with more detail, without headings.
4. Sections under level-2 headings, each a list of links in the form [name](url), optionally followed by a colon and a short note.

A section headed "Optional" marks links a tool can skip when it needs a shorter version.

For a small SaaS product, the file might contain lines like these:

• # Acme Invoices
• > Acme Invoices is invoicing software for freelancers: send invoices, take card payments and chase late payers automatically.
• ## Product
• - [Features](https://acme.example/features): what it does, page by page
• - [Pricing](https://acme.example/pricing): plans and limits
• ## Docs
• - [Getting started](https://acme.example/docs/start)
• - [API reference](https://acme.example/docs/api)
• ## Optional
• - [Changelog](https://acme.example/changelog)

Each of those would sit on its own line in the real file. Keep it short. The point is to choose, not to list everything.

Some sites also publish an llms-full.txt, a single file with the full text of their key pages. It's a common companion, not part of the original proposal.

Does llms.txt actually do anything?

Here the honest answer is: not much, so far.

In June 2026 Ahrefs published a study of 137,210 sites in its web analytics product that had traffic in May 2026. Ahrefs found that 28% of those sites published an llms.txt file, and 97% of those files received no requests at all that month. Search Engine Journal's write-up adds that AI retrieval bots made about 1.1% of the llms.txt requests that did happen. Ahrefs measured requests, not what any bot did with the file, and its customers lean technical, so adoption on the wider web is probably lower.

Google has said Search does not use llms.txt, and it plays no part in how Google ranks pages.

So why would anyone bother? A few reasons hold up:

• It costs very little. A useful file takes under an hour to write and rarely needs changing.
• Some developer tools support it. Support so far centers on documentation: some docs platforms generate the file, and some coding tools can read it. That makes it more useful for products with docs or an API.
• Writing it is a good exercise. If you can't summarize your product in one sentence and pick its five most important pages, your homepage probably has the same problem.

Treat it as cheap insurance, not a ranking lever. If you only have time for one thing, letting AI crawlers in and making sure your pages work without JavaScript matter far more. Our guide on how to show up in ChatGPT answers covers those steps.

How is llms.txt different from robots.txt and a sitemap?

They answer different questions:

• robots.txt says which crawlers may fetch which paths. It's a set of rules, and crawlers check it before they fetch anything.
• sitemap.xml lists every URL you want search engines to index, with optional last-modified dates. It's for completeness.
• llms.txt is a short, human-readable guide to the pages that matter most, written in Markdown. It's for focus.

llms.txt does not grant or deny access to anything. If your robots.txt blocks an AI crawler, an llms.txt file won't change that.

When is llms.txt worth having?

It's most worth adding when:

• Your product has documentation, an API or a developer audience, where support is strongest.
• Your site is large and a short guide to its key pages is genuinely helpful.
• You've already handled the basics: AI crawlers are allowed in, and your main pages render their content in the HTML.

It's least useful when your site is a handful of pages that are already easy to read, or when your real problem is that crawlers can't get in.

If you add one, serve it as plain text or Markdown at /llms.txt. A common mistake is a single-page app that answers every unknown path with its HTML shell, so /llms.txt returns a web page instead of the file.

How does MakerRadar use it?

MakerRadar publishes an llms.txt and an llms-full.txt that list every product, launch week, collection and guide on the site, and every page also answers a request for Markdown with a clean copy. We publish them because a directory's job is to be easy to read, not because we expect them alone to change anything.

Check it in a minute

These free tools need no sign-up:

• The llms.txt generator and validator checks whether /llms.txt exists and is served as text rather than your app's web page, looks for the title and summary line the spec asks for, checks the ## sections of links and flags links that don't parse, reports the file's size and whether /llms-full.txt exists, and can draft a file from your homepage's title, description and main links.
• The agent readiness checker looks for llms.txt and llms-full.txt alongside the other things AI agents look for: a Markdown answer to Accept: text/markdown, an MCP server card, an API catalog, an OpenAPI document, an agent skills index, Content Signals and AI crawler rules, and DNS-AID records.
• The AI visibility checker tests the step that matters more: whether your robots.txt lets 20+ AI crawlers in, and whether your page has readable text without JavaScript.

Building something? List it on MakerRadar and it appears in our own llms.txt too.