Free llms.txt Generator
We read your homepage, your sitemap and your key pages, and write the file from their real titles and descriptions.
Edit before downloading, or write it by hand
Enter your site and press Generate: you get a finished file built from your pages' own titles and descriptions. Download it and upload it to yourdomain.com/llms.txt. The output follows the llmstxt.org format: an H1, a one-line summary in a blockquote, H2 sections of described links, and an Optional section.
Honest caveat: no major assistant has committed to reading llms.txt, and Google says it does not use it. It is cheap and harmless, not a ranking lever.
Three sites per visitor per hour; each one reads up to about 60 pages. Editing and updating by hand is unlimited.
We downloaded the llms.txt of the 500 sites cited most often in 12,315 ChatGPT, Claude and Gemini answers we collected, on 28 September 2026. 202 serve a valid file. The median one is 10.8 KB, has 8 sections and 47 links.
Summary line: 175 of 202 open with a blockquote summary right under the H1. Descriptions: in the typical file, 90% of links carry one.
Optional: 61 files use the Optional section. No links at all: 24. Over 100 KB: 17, two of them over 1 MB.
A file on your site is the small half of AI visibility.
Assistants mostly answer from other people's pages: roundups, comparisons, niche blogs. MentionAgent finds the posts in your niche worth appearing in, emails the people who run them, and gets your product mentioned there. You approve every draft.
See how MentionAgent worksHow to generate an llms.txt file
- Generate. Enter your domain and press Generate. The tool takes the name and summary from your homepage, reads the sitemaps listed in your robots.txt (or
/sitemap.xml), and checks the usual key pages such as/pricingand/docseven if the sitemap leaves them out. It groups pages by the first part of their path, so/docs/...becomes a Docs section, and puts privacy and terms pages under Optional. Then it opens each chosen page and uses its own title and meta description. Tag archives, pagination, translations, login and thank-you pages are left out, and so are listing pages that share one generic title. - Trim it. Open Edit and delete sections a model does not need; a list of old press releases is not a map. A page with no meta description gets no description in the file, so add a few words after a second
|on the ones that matter, then press Update the file. - Write the summary like an answer. One sentence that says what you are and who it is for. It is the line most likely to be quoted.
- Generate, download, upload. Put the file at the root so it loads at
yourdomain.com/llms.txtas plain text, then run it through the llms.txt validator.
The llms.txt format
The proposal at llmstxt.org defines a Markdown file with a fixed order. Only the H1 is required; everything else is recommended.
| Part | Syntax | Required | What it is for |
|---|---|---|---|
| Name | # Site name | Yes | The first line of the file. |
| Summary | > One sentence | Recommended | What the site is, in the words you would want repeated. |
| Details | Plain paragraphs or lists | Optional | Context a model should read before the links. No headings. |
| Sections | ## Docs | Recommended | Groups of links, named for what they contain. |
| Links | - [Title](url): description | Recommended | One per line; the description after the colon is optional but common. |
| Optional | ## Optional | Optional | Links a tool can drop when it needs a shorter context. |
A minimal file that validates:
# Acme Analytics > Acme Analytics is product analytics for B2B SaaS teams: event tracking, funnels and retention without a data engineer. ## Docs - [Quickstart](https://example.com/docs/quickstart): Install and send the first event in five minutes - [API reference](https://example.com/docs/api): Every endpoint with examples ## Optional - [Changelog](https://example.com/changelog)
What 202 real llms.txt files look like
Formats on paper and files in the wild differ, so we checked the second. On 28 September 2026 we took the 500 domains cited most often across 12,315 answers from ChatGPT, Claude and Gemini that MentionAgent has collected, requested /llms.txt from each, and kept the 202 that returned a Markdown file starting with an H1. Among the 50 most-cited domains, 17 serve one.
The typical file is a short map, not a dump: a median of 10.8 KB, 8 sections and 47 links, with the middle half between 20 and 112 links. The conventions the proposal recommends are the ones people follow. 175 of the 202 put a blockquote summary directly under the H1, and among the 172 files that list links, the typical one gives 90% of its links a description after the colon. The exceptions are informative: 24 files contain no links at all, just a name and some prose, and 17 are over 100 KB, which is closer to llms-full.txt than to an index.
Section names follow what the sites sell rather than any standard. These are the most common H2 headings across the 202 files:
| Section heading | Files using it |
|---|---|
| Optional | 61 |
| Company | 30 |
| Pricing | 27 |
| Pages | 25 |
| Resources | 22 |
| Product | 20 |
| Blog | 18 |
| Posts | 17 |
| Contact | 17 |
| Solutions | 15 |
Pricing appearing in 27 files is the practical lesson: if a model is going to describe your product, the page it most needs is the one with the numbers on it. The generator puts a top-level /pricing page in the Pages section; keep it.
Does llms.txt actually work?
Nobody has shown that it changes what an assistant says about a site. It is a community proposal, not a standard; no major assistant has committed to reading it, and Google has said its search systems do not use it. When an assistant answers with a live search, it reads pages from the search results, not your llms.txt.
That is why the adoption numbers are the interesting part. The validator page has the longer version: about 40% of the sites assistants cite most serve a file, and the five most-cited domains (Reddit, Medium, YouTube, Facebook and the App Store) do not. Having one does not get you cited; being on the pages assistants read does. Treat the file as ten minutes of housekeeping and spend the next hour on the second part, which is what how to get cited by ChatGPT covers.
Frequently asked questions
How do I generate an llms.txt file?
Enter your site above and press Generate. The tool reads your homepage for the name and summary, your sitemaps for the pages, and each chosen page for its own title and description, then groups them into sections with legal pages under Optional. Trim anything a model does not need under Edit, download the file and upload it to yourdomain.com/llms.txt.
Does llms.txt actually work?
Nobody can show that it changes what an assistant says. It is a community proposal, no major assistant has committed to reading it, and Google has said its search systems do not use it. It is cheap to add and harmless, and 202 of the 500 sites cited most often in 12,315 ChatGPT, Claude and Gemini answers we collected serve one, but treat it as housekeeping, not a ranking lever.
What are the best practices for writing an llms.txt file?
Start with an H1 and a one-sentence summary in a blockquote (175 of the 202 real files we checked do). Group links under H2 sections named after what they are: Docs, Pricing, Blog. Give each link a short description after a colon; most files describe nearly every link. Keep it a map, not a copy of your sitemap: the median file has 47 links. Put pages a model can skip under an H2 called Optional.
Where do I put my llms.txt file?
At the root of your domain, so it loads at yourdomain.com/llms.txt, served as plain text. Upload it the same way you would robots.txt. Several WordPress SEO plugins can now publish one for you, so check your plugin's settings before adding a file by hand.
What's the difference between llms.txt and llms-full.txt?
llms.txt is an index: a name, a summary and links. llms-full.txt, from the same proposal, carries the full text of the pages concatenated into one Markdown file. It is much rarer and mostly used by documentation sites. This generator builds llms.txt.
What does the Optional section mean?
In the llmstxt.org format, an H2 named Optional marks links a tool can skip when it needs a shorter context. It is the one section name the proposal defines. 61 of the 202 files we checked use it.