llms.txt Generator

What is an llms.txt generator and what does this one do?

An llms.txt generator builds the file for your site root in the format of the llmstxt.org proposal. llms.txt is a short Markdown file that tells AI agents what your site is about and where the important pages live. Jeremy Howard published the proposal in September 2024. Version 2 arrived in August 2026 and made one rule clear: a file covers the pages under its own path.

This tool puts three jobs into one panel:

• Automatic: it reads your robots.txt and sitemap, groups up to 100 URLs into sections and takes the title and description of the first 50 pages from the pages themselves.
• Manual: you write the file from scratch in an empty editor, adding sections and links as you go.
• Validate: it checks your live file line by line, either from its URL or from pasted text.

On my own site I generate llms.txt in three languages, fresh on every request. It is also on the technical checklist I use in SEO consulting work. Still, I never treat it as a visibility lever on its own.

Does llms.txt actually do anything?

The honest answer depends on what you expect. Google is clear about this in its official documentation. You do not need new machine readable files, AI text files or special markup to appear in its AI features. In other words, llms.txt will not earn you Google rankings or AI Overviews visibility.

That said, the file has a real use. According to the 2026 notes of the proposal, thousands of sites publish one. Documentation platforms generate it automatically, and coding agents use these files reliably. In addition, Chrome Lighthouse looks for llms.txt in its agentic browsing audits. A missing file counts as not applicable, while Lighthouse flags a server error.

So here is my advice: treat llms.txt as a map that helps a visiting agent find the right page fast. The real work is still crawlable pages, clear content and trust signals. For that side of the story, see my guide to GEO.

How do you build a good file with an llms.txt generator?

The tool gives you a strong draft; your choices turn it into a good file. When I edit an llms.txt file, I follow this order:

1. Rewrite the summary. Most sites write the meta description to win clicks. The summary instead should say what the site does, who it serves and where it operates, in two sentences.
2. Prune the links. Not every URL in your sitemap belongs in the file. For most sites, services, product categories, pricing, the about page and core guides are enough; untick old campaigns.
3. Make descriptions informative. A concrete note beats a vague one. Google Ads setup, management and monthly reporting helps an agent far more than Our services.
4. Move secondary links to Optional. Privacy, cookie and terms pages can live there.
5. Use the Extra details field. Notes like service area, working language or booking rules go between the H1 and the sections as plain paragraphs.

Keep the file short. The proposal wants llms.txt small enough to fit in a context window, with the detail living behind the links. That is also why the tool warns you once the file passes 100 KB.

Where does llms.txt go on a multilingual site?

llms.txt does not have to sit at the root. Version 2 of the proposal says a file covers the pages under its own path. When several files apply, agents should use the most specific one. For example, /docs/llms.txt describes only the pages under /docs/.

This rule makes life easier on multilingual sites. Say your main language sits at the root, with English under /en/ and German under /de/. Then one file per language is the cleanest setup. The tool detects language prefixes and offers a Scope choice:

• Root: URLs without a language prefix; you upload the file to yourdomain.com/llms.txt.
• Prefixes such as en or de: only that language; the file goes into that folder, for example /en/llms.txt.
• All: every language in one file; however, this can make an agent work harder than necessary.

The tool shows the right upload location below the preview. For the links between your language versions, use the hreflang generator as well, because llms.txt does not do that job.

How do llms.txt, robots.txt and your sitemap fit together?

The three files have different readers, and none of them replaces another. Robots.txt tells crawlers which paths they may crawl. The sitemap gives search engines the full list of indexable URLs. An llms.txt file, on the other hand, hands an agent a curated summary.

Consistency matters here. Blocking an AI bot in robots.txt while guiding it with llms.txt sends a mixed signal. Therefore, review your rules with the robots.txt generator before you publish the file. The tool itself reads the Sitemap lines in robots.txt. Without them, it tries the common sitemap locations and then falls back to your homepage links.

If your sitemap is incomplete or stale, your llms.txt draft will be too. In that case, fix the list first with the XML sitemap generator and then run this tool. That way the draft reflects the real structure of your site.

What are Markdown versions and the describedby link for?

The 2026 version of the proposal describes two complementary steps. First, offer a clean Markdown version of pages an agent may need at the same URL, as page.html.md or page.md. Second, point to these files with standard link relations.

• rel="alternate" type="text/markdown": points to the Markdown version of a page.
• rel="describedby": points to the llms.txt file that covers the page.

You can add these as link elements in the HTML or as an HTTP Link header. The header route works through server or CDN settings, so you do not have to touch the pages. For most company sites and online stores these steps are optional. If you publish documentation, an API or a large knowledge base, though, Markdown versions make an agent's job much easier. I cover these steps in a wider frame in my article on technical SEO after AI.

How do you test the file after uploading it?

After you upload the file, switch to Validate mode and enter your domain. The tool fetches /llms.txt and checks these points:

• HTTP status: it should return 200. A 404 usually means the file sits in the wrong folder. After a redirect, the final URL must reach the right file.
• Content-Type: text/plain or text/markdown. Some servers answer a missing file with the homepage and a 200 code; the tool catches an HTML response separately.
• Encoding and size: UTF-8 and under 100 KB.
• Structure: one H1, a summary, ## sections, correctly formatted link lines and no repeated URLs.

If you use caching or a CDN, the old version may keep appearing for a while. When you suspect a redirect, check the address with the redirect checker too. Found errors? Press Open in editor, fix the file in Manual mode and download it again.