Free tool
Generate a spec-correct llms.txt in a minute
Fill in your name, one summary sentence and the pages worth reading. The file updates live, follows the llmstxt.org format, and never leaves your browser.
Free. Nothing is uploaded, no signup, no email. Updated August 24, 2026.
Your llms.txt
12 characters · save as /llms.txt at your site root
Fill in a name and a summary to enable copy and download.
Does anyone actually read llms.txt? The numbers
Before you spend an afternoon on this, here is what the measured evidence says as of August 2026. Short version: adoption is growing fast, use by AI engines is close to nil, and the file is cheap enough to publish anyway.
- 36,120
- sites with an llms.txt by May 2026
- Up from 4,088 in June 2025 — an 8.8× rise in twelve months, per Originality.ai’s tracker. Originality.ai, 2026
- 97%
- of llms.txt files got zero requests in May 2026
- Ahrefs analysed server logs from 137,000 domains. Of the requests that did arrive, GPTBot was 4.51% and ClaudeBot 0.80%. Ahrefs, June 2026
- 8.7%
- of the world’s top 1,000 websites publish one
- Adoption is concentrated in developer platforms and documentation sites that generate the file automatically. Rankability, June 2026
- 0
- real AI bots seen requesting it in a 900-domain audit
- An August 2025 log audit found no GPTBot, ClaudeBot or PerplexityBot requests for llms.txt at all — only auditing tools. Longato, August 2025
- “keywords meta tag”
- is what Google compared it to
- John Mueller: no AI services use it, and bots do not request the file. Search Engine Roundtable
What llms.txt is
llms.txt is a Markdown file at the root of a website that tells a language model, in a few hundred words, what the site is and which pages are worth reading. Jeremy Howard of Answer.AI proposed it on 3 September 2024; the spec reached v2 on 10 August 2026. The idea borrows from robots.txt and sitemaps: one predictable place a bot can look first, before it starts pulling HTML full of navigation, cookie banners and scripts.
The format is deliberately small. In order:
- An H1 with the site or project name — the only required line.
- A blockquote with a one-sentence summary that carries the key facts.
- Zero or more ordinary paragraphs with more detail.
- Zero or more H2 sections, each a Markdown list of
[Title](url): notelines pointing at the pages you want read. - An optional section headed Optional that an agent may skip when context is short.
A companion convention, llms-full.txt, concatenates the full text of those pages into one file so an agent can read everything in a single fetch. It makes sense for documentation; for a marketing site it mostly duplicates the pages.
What llms.txt is not (yet)
It is not a ranking signal, and it is not something the major AI engines have said they read. Google’s John Mueller said no AI system currently uses it and compared it to the keywords meta tag: a site owner’s claim about their own site, which a crawler has no reason to trust when it can read the site itself. Google’s guidance on generative AI features does not mention it at all.
The logs agree. In Ahrefs’ June 2026 study of 137,000 domains, 97% of llms.txt files received no requests in May 2026. Where requests did arrive, GPTBot accounted for 4.51% and ClaudeBot 0.80%; the biggest requesters were tools that check whether you have an llms.txt. An independent audit of about 900 domains in August 2025 saw no real AI bot request the file at all. OpenAI, Anthropic and Perplexity have published crawler documentation and none of it references llms.txt as of August 2026.
So why do 8.7% of the top 1,000 sites publish one, and why does the count keep growing? Because it costs ten minutes, it cannot hurt, and documentation-heavy companies find it useful for their own agents and customers’ agents. Cloudflare, Stripe, Vercel and Perplexity’s docs all serve one today. So do we: sightivo.com/llms.txt.
How to write a good one
- 01
One sentence that a stranger could repeat
The blockquote is the line an assistant is most likely to quote back. Category, who it is for, and the one thing that makes it different — the same sentence you would use in a directory listing, so every source says the same thing about you.
- 02
Ten pages, not a hundred
List the pages that answer the questions a buyer or an agent would ask: what it does, pricing, docs, comparisons, contact. Leave the blog archive out or put it under Optional. A file the length of a sitemap defeats the point.
- 03
Canonical, absolute URLs
One URL per page, the same one your canonical tag uses, with https and no tracking parameters. Relative paths are fine in the form above — we expand them with your homepage.
- 04
Notes that add a fact
The text after the colon should say what is on the page, not repeat the title. "Plans and limits, updated monthly" beats "Pricing page".
- 05
Keep it current
A stale llms.txt is worse than none, because it is the one document you told machines to trust. Revisit it whenever pricing, positioning or the product name changes.
- 06
Fix robots.txt first
The file that actually decides whether ChatGPT, Claude or Perplexity can read you is robots.txt. Make sure GPTBot, OAI-SearchBot, ClaudeBot, Claude-SearchBot and PerplexityBot are allowed before you worry about a file none of them fetch.
Where to put it, and how to test it
Save the download as llms.txt in whatever folder becomes your site root — public/ on most static frameworks — and deploy. It should then load at https://yourdomain.com/llms.txt as plain text, first line starting with #. If you also want llms-full.txt, paste the Markdown of your key pages under matching headings in a second file.
To test whether anyone reads it, do not use a checker tool — they are most of the traffic these files get. Look at your server or CDN logs for the path over a month. If it is quiet, that is the truth about llms.txt today, and it costs you nothing.
Should you bother? Yes, because it takes ten minutes and the discipline of writing the summary sentence is worth more than the file. No, if you expected it to change what ChatGPT says about you. That is decided by the sources ChatGPT actually cites — and you can check what it says about you for free.
llms.txt questions
- What is llms.txt?
- A plain Markdown file at /llms.txt that describes a site for AI systems: an H1 with the name, a one-line summary in a blockquote, optional paragraphs, and H2 sections listing the pages worth reading as Markdown links. Jeremy Howard of Answer.AI proposed it on 3 September 2024; the spec at llmstxt.org was revised to v2 on 10 August 2026.
- Does Google use llms.txt?
- No. Google has said no AI system currently uses it, and John Mueller compared it to the old keywords meta tag — a site owner’s claim about their own site rather than something a crawler needs. Google’s own guidance on generative AI features says there is nothing special to add beyond normal SEO.
- Does ChatGPT read llms.txt?
- There is no confirmation from OpenAI, and the server-log evidence says it is rare. Ahrefs’ June 2026 study of 137,000 domains found 97% of llms.txt files received zero requests in May 2026; of the requests that did arrive, GPTBot accounted for about 4.5% and ClaudeBot under 1%, with most coming from tools that audit whether the file exists.
- Where do I put llms.txt?
- At the root of your domain, so it is reachable at https://yourdomain.com/llms.txt. The spec also allows one at any sub-path to describe the pages beneath it (for example /docs/llms.txt). Serve it as plain text with UTF-8 encoding.
- What is the difference between llms.txt and robots.txt?
- robots.txt tells crawlers what they may not fetch and is honoured by every major bot, including GPTBot, OAI-SearchBot, ClaudeBot and PerplexityBot. llms.txt is the opposite kind of file: a suggestion of what is worth reading, honoured by nobody in particular yet. If you only do one, make sure robots.txt is not blocking the AI crawlers you want.
- What is llms-full.txt?
- A companion convention where the full text of your key pages is concatenated into one Markdown file, so an agent can read everything in a single fetch instead of following links. It is useful for documentation sites with many short pages and mostly pointless for marketing sites.
- Does llms.txt help SEO?
- Not directly. It is not a ranking signal for Google, Bing, or any AI engine that has published guidance. What helps is the work the file makes you do: writing one clear sentence about what you are, choosing the ten pages that matter, and making sure each has a canonical URL. That clarity shows up in your pages, which is what engines actually read.
- How do I check if my llms.txt works?
- Open https://yourdomain.com/llms.txt in a browser: it should load as plain text with a # heading on the first line. Then check your server logs or CDN analytics for requests to that path; if none arrive over a month, that is the honest answer about whether anyone reads it. Our own file is at sightivo.com/llms.txt if you want a working example.