What this does, and what it deliberately does not

The generator reads your public sitemap, groups your URLs by their path structure, and gives you a properly formatted llms.txt draft with the sections already laid out.

It does not write your descriptions. Every one is marked REPLACE THIS, and that is the point. A sitemap tells you a page exists at a URL. It cannot tell you what that page answers, which is the only genuinely useful information in the file.

We could have generated plausible descriptions from page titles. It would demo better and it would produce a file that describes your business slightly wrongly in a place designed to be read by machines. That is worse than an obvious placeholder.

The honest position on llms.txt

It is a convention, not a standard. No AI system is obliged to read it, and the evidence that any particular one does is thinner than the enthusiasm around it suggests.

So why publish one? Because the cost is close to zero and the downside is nil. If a system does read it, you have handed it a clean map of your site and a clear statement of what you do. If none does, you have spent an hour and lost nothing.

What it will not do is make you visible. If a model cannot resolve who you are, or your pages carry no extractable answer, an index file does not fix either problem. Anyone selling llms.txt as the answer to AI visibility is selling the easiest item on the list as though it were the whole list. The parts that decide citations are entity authority, extractable content, and third-party corroboration. This file is genuinely useful housekeeping alongside them.

How to finish the draft

  1. Write the opening summary. One or two sentences a model could quote to describe you.
  2. Add the context paragraph. What you sell, who to, what makes you different, and any facts you actively want quoted, such as pricing or coverage. If you publish prices, put them here.
  3. Describe each page in one line, saying what it answers rather than what it is called. “Pricing for regulated firms, with the compliance layer explained” beats “Pricing”.
  4. Cut ruthlessly. Keep it under 100 links. An index of everything is an index of nothing, and a long file dilutes the pages you actually want cited.
  5. Publish at the root and check it loads at yourdomain.com/llms.txt.
  6. Review it when your services or prices change. A stale fact file is worse than none, because it is confidently wrong.

Ours is at margen.net/llms.txt if you want to see a finished one, and /llm-info.html is the companion piece: the same facts written for machines in more depth.

The limits of this tool

It reads your sitemap and nothing else. It cannot tell whether the pages in it are any good, whether they are extractable, whether your sitemap is complete, or whether the pages you most want cited are even in it. It caches by domain for 24 hours, so if you change your sitemap and re-run immediately you will see the previous result.

If your sitemap is missing, stale or partial, the draft inherits all of that. That is worth checking before you trust the output.

llms.txt Generator: Common Questions

What is llms.txt and do I need one?

An llms.txt file is a plain-text index at the root of your domain that tells AI systems what your site contains and what each page is for. It is a convention rather than a standard anyone enforces, so no system is obliged to read it. It is cheap to publish, it costs nothing to be wrong about, and it makes your site easier to understand for anything that does read it.

Will publishing an llms.txt file get me cited by AI?

On its own, no. It makes your site easier to navigate for a system that reads it, which is worth having, but it does not create the entity clarity or third-party corroboration that decides whether a model names you. Treat it as one cheap item on a longer list rather than the thing that fixes AI visibility.

Why does the generated file say REPLACE THIS everywhere?

Because the useful part of an llms.txt file is the human judgement, and a sitemap cannot supply it. The generator gives you the structure, the grouping and the URLs. What each page actually answers, and what you want a model to know about your business, has to be written by someone who knows the answer.

Where do I put the file?

At the root of your domain, so it sits at yourdomain.com/llms.txt. Not in a subfolder. If your site is on a platform that will not let you add a root-level text file, that is worth knowing, because the same constraint usually affects robots.txt and verification files too.

Do I have to give you an email address?

Not to generate it, and not to copy it. The generator runs and the copy button works with no email at all. We ask for one only if you want the file downloaded rather than copied, which is an honest trade rather than a wall.

What if the tool cannot find my sitemap?

Point it directly at the XML file, for example https://example.com/sitemap.xml. The tool checks your robots.txt and the two usual locations first, but plenty of sites keep their sitemap somewhere else. If you do not have one at all, that is a bigger finding than the llms.txt question and worth fixing first.

Get the version a person does

This builds the file. It cannot tell you whether AI systems can resolve your business, whether your content is extractable, or who is being cited instead of you.

Run the free AI Visibility Score for the automated version, or get in touch for the manual one.