Free LLMs.txt Generator
Enter your domain and get a spec-compliant llms.txt back in a couple of seconds. It reads your sitemap where you have one and crawls where you do not, and it uses the real title and description already on each page rather than writing new ones. Copy it, save it at your root, done. No signup, nothing stored.
LLMs.txt Generator
Generate an llms.txt file from your site's sitemap (or a homepage crawl if none is found), with a real title and description fetched from each page. Takes 20-40 seconds, not instant, since it visits each page individually.
Want to know if assistants mention your brand at all before you write this file? Try the AI Visibility Tool.
Already have an llms.txt live? Check it against the spec with the LLMs.txt Validator.
What llms.txt is
llms.txt is a plain markdown file at the root of your domain that tells a language model what your site is and which pages matter. It was proposed by Jeremy Howard of Answer.AI in September 2024, and the idea is simple: a model reading your site has to strip navigation, adverts and boilerplate out of your HTML to find the substance. This file hands it the substance directly.
The format is a heading with your site name, a blockquote summarising what you do, then sections of links with a one-line description each.
Who actually reads llms.txt today
Being straight, because most pages on this topic are not: adoption is partial, and no major assistant has publicly committed to using llms.txt as a ranking or citation input. Some crawlers fetch it. A number of developer-tooling companies publish one. It is a proposal with real momentum, not a standard.
So the honest case for publishing one is not that it will get you cited tomorrow. It is that the file takes fifteen minutes to produce, costs nothing to host, breaks nothing if it is ignored, and puts you in the set of sites that are ready if adoption does arrive. That is a reasonable bet at that price. Anyone telling you it is essential is overselling it.
What it will not do: replace your sitemap, affect your Google rankings, or stop a model that was going to ignore you from ignoring you.
The format, section by section
One H1 at the top with the site or company name. Required - a file that opens with anything else fails most validators.
A blockquote immediately after, summarising what the site is in a sentence or two. This is the single highest-value line in the file, because it is the description a model is most likely to reuse verbatim.
H2 sections grouping your links - Pages, Docs, Products, whatever fits. Optional but expected.
Links as markdown list items in the form hyphen, name in square brackets, URL in parentheses, colon, then a one-line description. Bare URLs do not parse and are the most common formatting failure.
An optional section for links a model can skip. Useful on large documentation sites, unnecessary on most others.
llms.txt, llms-full.txt, robots.txt and sitemap.xml
Four files at your root, four different jobs, and they are routinely confused.
robots.txt says which paths a crawler may request. It is permission.
sitemap.xml lists every URL you want discovered. It is inventory.
llms.txt says what your site is about and which pages matter, in prose a model can read. It is a summary.
llms-full.txt holds the actual content of those pages as markdown, so a model does not have to fetch them individually. It is the full text.
Publish robots.txt and sitemap.xml always. Publish llms.txt if you want to be described accurately. Publish llms-full.txt only if your content is genuinely reference material that people ask assistants about - documentation, APIs, technical guides. For a marketing site it is overkill.
Why this tool uses your existing descriptions
Every other generator on this SERP writes the descriptions with a language model. This one does not, and the difference is the point of the file.
llms.txt exists so an assistant can describe you accurately. If the descriptions in it were themselves generated - polished, optimistic, subtly different from what your pages actually say - then the file is a second version of your site rather than a map of it, and the one thing it was supposed to fix is the thing it introduces.
So this tool pulls the title and meta description already on each page. If those are weak, the file will be weak, and the fix is to improve the pages rather than to have a model paper over them. The Meta Tag Generator is the tool for that, and running it first is a sensible order of operations.
Where to publish it and how to keep it current
Save the file as llms.txt at the root of your domain, so it sits at yourdomain.com/llms.txt, alongside robots.txt. It must be served as plain text - a platform that serves it as HTML with a download prompt has broken it.
On Shopify and Framer, where you cannot write arbitrary files to the root, you may need a redirect or a hosted route. Check that the file loads as text in a browser before assuming it worked.
Regenerate when your site structure changes - a new section, renamed pages, a launch. Monthly is more than enough for most sites, and there is no benefit to updating it on a schedule if nothing has moved.
Need this built out properly, sitewide? See our Generative Engine Optimization services.
FAQ
Frequently Asked Questions
Do AI assistants actually read llms.txt?
Some crawlers fetch it. None of the major assistants has committed to using it as a ranking or citation input. It is a proposal with real momentum rather than a standard. It takes fifteen minutes to publish and costs nothing if it never catches on.
Where does the file go?
At the root of your domain, at example.com/llms.txt, exactly like robots.txt.
What is the difference between llms.txt and llms-full.txt?
llms.txt is a map - titles, links and one-line descriptions. llms-full.txt holds the actual page content in markdown. Publish llms.txt first, and llms-full.txt only if your content is genuinely reference material.
Does this replace my sitemap?
No. A sitemap tells search crawlers which URLs exist. llms.txt tells a language model what your site is about and which pages matter, in a format it can read without parsing your HTML.
Why does this tool use my real page descriptions instead of writing better ones?
Because the file exists so an assistant can describe you accurately. A generated description that flatters your page is exactly the thing that makes the file untrustworthy.
How often should I regenerate it?
Whenever your site structure changes - new sections, renamed pages, a launch. Monthly is plenty for most sites.