llms.txt Generator
Generate optimized llms.txt files with validation and installation guide.
llms.txt Generator that reads your real sitemap
Paste your domain. We follow your robots.txt to your sitemap index, pull the titles and descriptions your pages actually carry, and hand back a file that matches the v2 spec. No invented URLs, no placeholder links, no signup. And because you will ask: we also tell you what the spec's author and Google have said about whether anything reads it yet.
Built and reviewed by Sayed Hasan, founder of SEOs Hut · Updated 27 September 2026 · Free, no signup
What is llms.txt, and does anything actually read it?
llms.txt is a Markdown file at the root of your site that lists your most important pages with a short description of each, so an AI assistant can find your key content without crawling everything. It was proposed by Jeremy Howard in September 2024 and reached v2 in August 2026. No AI company has published a statement that its consumer product reads yours, and Google says its systems ignore the file.
Before you publish: llms.txt is not "robots.txt for AI"
That analogy is on nearly every page that ranks for this term and it is wrong in both directions. robots.txt is a ratified access-control standard that crawlers obey. llms.txt is an unratified content-summary proposal with no enforcement, no permission semantics, and no confirmed reader among the consumer AI products. llmstxt.org says plainly that the two files have different purposes. If what you want is to control which AI bots can fetch your site, that is a robots.txt job — check it with the AI Crawler Access Checker.
What the generated file actually contains
Everything a valid llms.txt can hold, in the order it appears, with the v2 status of each part. Most of it is optional — a fact none of the other generators mention.
An H1 with the name of the project or site. Per the spec this is the only required section — a file containing nothing but this line is valid.
A blockquote summary. One or two sentences describing what the site is. Optional, but it is the line an assistant is most likely to lift verbatim, so it is worth writing properly.
Free Markdown paragraphs. Anything that adds context an assistant would otherwise have to infer. Keep it factual — this is not a place for taglines.
An H2 opens a file list. The heading is your label, not a keyword from the spec, so name it whatever the group genuinely is.
A Markdown link with an optional description after the colon. This is the workhorse line, and it is where our generator uses your page's real title and meta description rather than inventing one.
In v1 an H2 literally named "Optional" marked pages that context-expansion tooling could skip. v2 removed that mechanical meaning along with the
llms_txt2ctx tooling. If your file still leans on it, the heading is now just a heading.v2 added discoverability through standard link relations —
rel="describedby" for the covering llms.txt, rel="alternate" with type="text/markdown" for a page's own Markdown version. Deliverable as an HTML element or an HTTP Link: header. This goes in your pages, not in the file.v2 also accepts both Markdown URL forms (page.html.md and page.md) and defines subpath scoping: a file covers the pages under its path, and the most specific file wins.
Google has answered this question twice, in public
“FWIW no AI system currently uses llms.txt. It's super-obvious if you look at your server logs — none of them fetch the llms.txt file.”
<b>John Mueller, Google, June 2025</b> — Search Engine Roundtable
Asked again almost a year later, in June 2026, he had not moved: "I don't think anyone knows — it's purely speculative for now (the file has existed for years, yet none of the AI systems use it — what does it mean?)". Google Search Central puts it in its own documentation: you do not need to create new machine-readable files, AI text files or markup to appear in AI Overviews or AI Mode.
We are telling you this on the page where we give you the generator, which is an odd sales technique, so here is the reasoning. Every page that ranks for this term promises the file will get you cited and none of them cite a source, because there isn't one. If we repeated that claim you would have no reason to trust anything else we publish. The honest case for publishing the file is smaller and still real, and it is in the next section.
Three reasons to publish it anyway
None of these is "it will get you into ChatGPT". All three are things the file does today, regardless of whether a consumer assistant ever fetches it.
It costs about ten minutes
A static text file at your root. No plugin, no build step, no ongoing maintenance beyond adding a line when you publish something important. The expected value only has to be slightly above zero to be worth it.
Writing it is a content audit
You cannot list your twenty most important pages without discovering that four of them are thin, two are duplicates and one redirects. That exercise is the real payoff, and it is why we make you look at the list before you upload.
The tooling side already uses it
OpenAI, Anthropic and Google all publish llms.txt for their own developer documentation, and coding agents genuinely consume those files. If your site has docs an agent might read, this is the audience that exists today.
Everything that has actually happened to llms.txt
Dated, sourced, and short — because the length of this list is itself the answer to "is this an established standard?"
Jeremy Howard proposes the file
Published at llmstxt.org by the Answer.AI co-founder. A proposal, not a standard — no working group, no RFC, no registry.
Google says nothing reads it
John Mueller, on the record: no AI system uses llms.txt, and server logs show none of them fetching it.
Lighthouse adds an audit — with a caveat
Chrome's Lighthouse gained an llms.txt audit under agentic browsing. It only flags the file if the server errors; a 404 is marked Not Applicable because "providing the file is optional at the moment". This is routinely mis-reported as Google endorsing the file.
Google says it again
Mueller calls it "purely speculative for now" and names WebMCP as the direction he prefers.
v2 of the spec lands
Adds discoverability through link relations, allows both Markdown URL forms, defines subpath scoping, and strips the special meaning from the Optional section.
Still zero published consumers
No AI vendor has published a statement that its consumer product reads third-party llms.txt files. Several publish one for their own docs. Those are different claims.
The four ways people get this file wrong
All four come from treating llms.txt as a ranking lever rather than a table of contents.
Listing every URL in the sitemap
What happens: A 400-line file that reproduces your sitemap adds nothing an assistant could not already get, and buries the eight pages that matter. The whole premise of the file is selection.
Do this instead: Cap it at the pages a human would send a new customer. For most small sites that is between ten and thirty lines. Our generator demotes legal pages and pagination for exactly this reason, but you should still cut the list further by hand.
Writing marketing copy in the descriptions
What happens: "Your trusted partner for excellence in plumbing" tells an assistant nothing it can use, and if anything ever does read the file, that is the sentence it has to work with.
Do this instead: Describe what is on the page in the plainest terms available. "Boiler repair prices and the makes we cover" beats any adjective you were considering.
Publishing it and calling AI visibility done
What happens: The file is the cheapest thing on the list and the least likely to be read. Meanwhile the site may be blocking the fetchers that genuinely matter, which no amount of Markdown fixes.
Do this instead: Check access first with the AI Crawler Access Checker, then check whether your pages are structured to be quoted with the AEO Readiness Checker. The file is step three, not step one.
Never looking at whether it is fetched
What happens: People publish the file, assume it is working, and have no evidence either way for a year.
Do this instead: Grep your access log for /llms.txt. One line tells you more than every blog post on this subject. If you want that read for you, the AI Crawler Log Analyzer separates the AI fetchers from everything else.
The five words this topic keeps confusing
- llms.txt
- A Markdown table of contents for your site, at the root. Descriptive. Nothing enforces it and no consumer AI product has published that it reads yours.
- robots.txt
- A ratified standard (RFC 9309) that tells crawlers what they may fetch. Prescriptive, and the major AI crawlers do honour it — with documented exceptions for user-initiated fetchers.
- llms-full.txt
- An informal companion convention holding the full text of your content rather than links to it. It is not in the v2 spec. Treat it as a community idea, not a requirement.
- Markdown alternate
- A plain-Markdown version of a single HTML page, linked with
rel="alternate" type="text/markdown". New in v2, and a separate thing from the llms.txt file itself.
- Grounding
- When an AI answer is built from live retrieved pages rather than the model's memory. Being retrieved at grounding time is what actually gets you quoted — and it runs on crawlability and indexing, not on this file.
If you only do one thing after this page
Publish the file, then check your logs in a fortnight. If nothing has fetched it, you have lost ten minutes and gained an audited list of your best pages. If something has, you will know before anyone writing about this topic does.
llms.txt Generator FAQ
Does ChatGPT read llms.txt?
Is llms.txt an official standard?
Where do I put the llms.txt file?
Will llms.txt improve my rankings in Google?
What changed in llms.txt v2?
What is the difference between llms.txt and llms-full.txt?
How many pages should I list?
Primary sources used on this page
- The /llms.txt file, v2 — the specification itself — llmstxt.org
- llms.txt changelog: what v2 added and removed — llmstxt.org
- Google's John Mueller: no AI system currently uses llms.txt — www.seroundtable.com
- Google: llms.txt is "purely speculative for now" (June 2026) — www.searchenginejournal.com
- Google Search Central: AI features — no new files or markup needed — developers.google.com
- Chrome Lighthouse: the llms.txt audit and why a 404 is Not Applicable — developer.chrome.com
Check the things that actually gate AI visibility
llms.txt is the cheapest item on the list and the least proven. These four are the ones with documented effects, roughly in the order worth doing them.
Read the guide: Does llms.txt Work? What Google and AI Bots Really Do With It and Robots.txt Best Practices: 10 Rules for SEO. Want it handled for you? See our AI SEO (GEO) services.