llms.txt: what it is, who actually uses it, and how to publish one
llms.txt is a proposed Markdown standard that gives large language models a curated map of a site’s most valuable content. It sits at yourdomain.com/llms.txt, alongside robots.txt et sitemap.xml. As of May 2026, no major AI provider has confirmed it parses the file — but it’s risk-free to publish and may earn a small early-adopter edge if the proposal is ever standardized.
Use the free generator below to create your own llms.txt in under five minutes, then keep reading for the full picture — how it works, who supports it, where it fits next to robots.txt and sitemap.xml, and the honest take on whether it moves the needle for SEO, AEO, or GEO.
What llms.txt is — and what it isn’t
llms.txt is a proposed standard, similar in spirit to robots.txt and sitemap.xml, but aimed at language models rather than search crawlers. Where robots.txt controls what crawlers may access and sitemap.xml lists every URL on a site, llms.txt tells an AI which handful of pages best represent the site and links to clean, Markdown-friendly versions of them.
What it isn’t: an official, enforced standard. No browser or AI assistant is required to read it. Treat it as a polite, structured introduction to your site for any LLM that chooses to use it.
How the file works (with a real example)
Un llms.txt file is plain Markdown with a simple shape: a top-level # heading naming your site, an optional blockquote with a one-line description, then one or more ## sections that group links to the most important pages.
# Riman Agency
> AI & SEO consultancy helping brands win citations in ChatGPT, Google AI Overviews and Perplexity.
## Core services
- [Answer Engine Optimization](https://rimanagency.com/services/aeo): How we make sites the most-cited answer.
- [SEO audits](https://rimanagency.com/services/seo): Technical-first audits that fix crawl, index, and content.
## Resources
- [llms.txt generator](https://rimanagency.com/llms-txt-generator): Free tool to build your own file in 5 minutes.
- [Blog](https://rimanagency.com/blog): Long-form notes on AEO, GEO, and AI search.
## Contact
- [Get in touch](https://rimanagency.com/get-in-touch): Send us a brief.
That’s the whole file. Some implementations also support llms-full.txt (an expanded, full-content version) and per-page Markdown alternates (e.g. /about.md) so an LLM can fetch the article body without the surrounding HTML.
How to create your own llms.txt in five steps
- Start a plain Markdown file nommé
llms.txt. Any text editor will do. - Ajouter un
#H1 with your business name, followed by a one- or two-line description in a>blockquote. - Group your most important pages under
##headings — for example: Core services, Pricing, About, Contact. Each item is a Markdown link:- [Page name](https://...), optionally followed by a short description. - Link to clean, content-rich URLs. If you can publish Markdown versions of those pages (e.g.
/services/aeo.md), even better — LLMs prefer the cleanest text they can find. - Host the file at your root domain:
https://yourdomain.com/llms.txt. Keep it updated whenever your important pages change.
If you’d rather skip the manual write-up, the generator at the top of this page takes you through all five steps with a guided form.
Adoption status: who actually supports llms.txt?
Short answer: no major AI provider has publicly confirmed that it reads llms.txt. Some publish one for their own product; none have committed to honoring it on the sites they crawl. Status as of May 2026:
| Fournisseur | Publishes its own llms.txt? | Crawler honors third-party llms.txt? |
|---|---|---|
| OpenAI (ChatGPT, SearchGPT) | Non | No public commitment; not observed in logs. |
| Anthropic (Claude) | Yes — docs.anthropic.com | Has not confirmed its crawler uses the file. |
| Google (Gemini, AI Mode) | Non | No commitment; included as a referenceable format in the A2A protocol (April 2025). |
| Meta AI | Non | No commitment. |
| Perplexité | Non | No commitment. |
That doesn’t make the file useless — it makes it speculative. Publish it because it’s a 10-minute job with no downside, not because there’s evidence of an AI traffic lift.
Companies that already publish an llms.txt
A growing handful of dev-tool and AI-native companies publish one as a soft endorsement of the format. Notable examples:
- Mintlify — auto-generates
llms.txtfor its documentation customers. - Tinybird — clean, well-structured example covering product docs.
- Cloudflare — used across the developer documentation portal.
- Anthropic — publishes one at docs.anthropic.com.
For a continuously-updated registry of sites that publish the file, browse directory.llmstxt.cloud.
The honest take: low effort, no risk, unproven benefit
The most-cited critic in the SEO community is Google’s John Mueller, who summarized the skeptical view bluntly:
“None of the AI services have said they’re using LLMs.TXT — and server logs show they’re not even checking for it. It’s comparable to the keywords meta tag: this is what the site owner claims their site is about. Why not just check the site directly?”
That’s the right frame. Treat llms.txt as an early-stage hint, not a ranking signal. It costs almost nothing to publish, breaks nothing if you do, and may pay off if a major provider eventually adopts the spec.
How llms.txt fits SEO, AEO, and GEO
It’s tempting to file llms.txt under “AI search optimization” and call it a tactic. It isn’t — at least not yet. To stay honest about where it actually lives, place it inside the three disciplines this site teaches:
SEO — the eligibility layer
SEO earns admissibilité: crawlable, indexable, technically healthy pages that match query intent. llms.txt is not a ranking factor. If your page can’t be crawled, indexed, or rendered, an llms.txt file does nothing. Treat it as a technical-SEO curiosity — like an XML sitemap variant — not a ranking lever.
AEO — earning selection
Answer Engine Optimization is the real play. AEO is the work of becoming the most reference-worthy answer across Google AI Overviews, Google AI Mode, ChatGPT Search, Perplexity and Claude. Pages win citations through the Triangle de citation — clarity, evidence, entities and structure — not through a Markdown index file. The way to win is exactly what this page demonstrates: a direct answer module up top, an adoption table that’s easy to lift, a working code example, a clearly attributed quote, and an FAQ block that maps to the questions users actually ask.
GEO — being represented accurately inside generated answers
Generative Engine Optimization is the broader practice of being represented correctly when an AI synthesizes an answer from many sources. Wins come from: stating facts unambiguously; attributing claims to named sources; keeping a visible “last updated” date; and structuring content so it’s easy to extract. llms.txt is often pitched as a GEO tactic — but there’s no evidence engines use it. Real GEO gains come from authoritative, well-structured, frequently-cited content that LLMs can reliably summarize without distortion.
Bottom line: publish an llms.txt, but don’t expect it to move metrics. Treat the page hosting it — like the one you’re reading — as the real SEO/AEO/GEO asset: answer-first, evidence-backed, structured, and entity-rich.
Foire aux questions
What is llms.txt?
llms.txt is a proposed Markdown standard that gives large language models a curated index of a site’s most valuable content. The file sits at https://yourdomain.com/llms.txt and lists the pages an LLM should look at first.
Do ChatGPT, Claude, or Gemini actually use llms.txt?
As of May 2026, no major AI provider has publicly committed to parsing the file. Server-log analyses show their crawlers don’t routinely request it. Treat it as a hint, not a contract.
How is llms.txt different from robots.txt and sitemap.xml?
robots.txt controls what crawlers may access; sitemap.xml lists every URL for search engines; llms.txt is an unofficial, curated content map aimed at LLMs.
How do I create an llms.txt file?
Write a Markdown file with an # H1 for your site, group your most important pages under ## headings as Markdown links, host it at your root domain, and keep it updated. The generator at the top of this page does this for you in five minutes.
Where do I host the file?
At your root: https://yourdomain.com/llms.txt. Anywhere else, and crawlers won’t find it.
Is llms.txt worth it in 2026?
There’s no measurable AI-search or referral lift attributable to llms.txt today, but no downside either. Publish it for the possible early-adopter edge — not because it will move traffic this quarter.
Will llms.txt help my SEO or AI search rankings?
No. There is no proven ranking or visibility benefit. Pages still win citations through the same fundamentals: clear answers, structured evidence, entity coverage and authority signals.
Which companies already publish an llms.txt?
Mintlify, Tinybird, Cloudflare and Anthropic, among others. A continuously-updated registry lives at directory.llmstxt.cloud.
Can I add metadata like token budgets or page priorities?
Some developers experiment with extra metadata in the file. There’s currently no evidence that crawlers parse or honor it — stick to the standard format until that changes.
How do I actually measure AI bot activity if llms.txt doesn’t help?
Use server-side AI-bot tracking (tools like Bot Analytics or Cloudflare’s AI Audit) to see which assistants are fetching your pages, and brand-mention tools (Brand Radar, Profound, AthenaHQ) to see how AI answers describe your business.
Sources and how we know
- llmstxt.org — the original specification by Jeremy Howard.
- Ahrefs — What is llms.txt? (adoption status, John Mueller commentary).
- SE Ranking — llms.txt complete guide (format, examples, criticism).
- directory.llmstxt.cloud — public registry of sites publishing the file.
- Tarek Riman, Intro to Answer Engine Optimization — From SEO to AEO (2nd Edition) — the AEO / GEO / SEO framework used in the section above.
Dernière mise à jour : May 30, 2026 · Editor: Tarek Riman.