llms.txt generator

llms.txt is the curated map AI crawlers read to understand your site - what you are and which pages matter. Fill in the fields and your file builds live below; copy or download it when the preview looks right.

  • No signup required
  • Copy or download llms.txt
  • Publishes at your root
Pages the AI should read

Pick your strongest entry points: the homepage, docs, pricing, and your best comparison or how-to pages.

llms.txt · live preview
# Your brand

## Pages worth reading

- [Home](https://example.com/): What we do, in two sentences.

Save it as llms.txt in your site root so it serves at https://yourdomain.com/llms.txt.

A good map gets you read.Sonarvue then tracks what the engines do with you - mentions, citations, share of voice.
Start the trial

What an llms.txt file is

An llms.txt file is plain markdown published at the root of your domain, at https://yourdomain.com/llms.txt. The file answers an ongoing question about how LLM-driven tools should get a clean, opinionated view of a site: sitemaps list everything, and they are undifferentiated - a /thank-you page and your best comparison post carry equal weight. llms.txt is the curated map. An H1 with your brand, a one-line summary that leads with what you are, then links to the handful of pages that actually explain the business. AI engines that request the file (and agents reading your site) get orientation in one fetch.

What this llms.txt generator builds

The generator produces the format the spec proposes and real-world examples follow: a single H1 (never two), a blockquote summary, your site URL as a deduplicated fetch target, a Pages worth reading section of markdown links with optional one-line notes, and an optional Contact section. Every field is live: change a page title or a note and the preview rebuilds instantly. The notes matter more than they look - an engine skimming the summary gets your positioning plus the purpose of each URL without having to open anything. Keep titles short, keep notes in the same voice as your site, and keep the total page count reasonable: a dozen strong links outperform sixty vague ones.

One practical caution this generator can't judge for you: URLs go in exactly as they serve. Lowercase, canonical, trailing behaviour matching what actually loads. A link that 404s is worse than a link that was never listed.

Which pages earn a place in the file

Pick pages the way a new hire would orient: homepage (or a product page that doubles as one), documentation and how-tos, pricing, and your two or three best comparison or “answer-the-buying-question” posts. Skip blog listings, tag pages, your careers page and anything with a session calendar - llms.txt is a map for answer engines, not a sitemap clone. If a page would not survive being quoted, it does not belong in the section.

How AI engines actually use it today

Honest version, matching what our answer-readiness checker reports: most engines do not request llms.txt yet, Google ignores it entirely, and its weight in the score is deliberately small. Its value is being cheap to create, easy to maintain as a source of truth as a source of truth for “the pages that matter”, and increasingly relevant as agent-driven browsing looks for exactly this kind of hand-off. Publish it, keep the summary in step with your positioning, re-run the readiness check afterwards, and move on to better-earned work. Then track what engines say about you over time - the trial scan is the natural next step.

Publishing the file in four steps

Generate above → download llms.txt → drop it in the root that serves your domain (the public/ folder on most static hosts, Uploads + a redirect on WordPress) → verify that visiting https://yourdomain.com/llms.txt renders raw markdown, then confirm with a scan that the readiness score reflects it. Pair it with an AI crawler robots.txt so the bots that would read the file are also permitted to arrive in the first place.

Keeping the file worth its place

A llms.txt that drifts out of sync with the site is worse than no file at all: the map now lies by omission, and an engine that trusted it wastes its fetches while your strongest pages stay undiscovered. Review the file on the rhythm your marketing already has - whenever you ship a product page, a pricing change or a good comparison post, regenerate, re-download and re-publish in the same sitting. The generator is deliberately fast so the refresh is a two-minute chore, not a project. Keep the summary human-written and short; delete a link rather than stretch its note when the page's title already says enough. And if you cannot say, at a glance, why a URL is in the list - that is the signal its time is up.

Frequently asked questions

What is an llms.txt file?

llms.txt is a plain-markdown file at the root of your domain that gives AI crawlers a curated, machine-readable map of your site: what you are, a one-line summary, and the specific pages worth reading. It is to AI engines what a sitemap is to search engines, only with opinions.

Does Google search use llms.txt?

No useful amount of Googlebot indexes llms.txt, and Google has given no signal that it will. It matters for AI answer engines and agents, some of which fetch it opportunistically. Treat it as a cheap, low-effort addition to a site that is already readable - the checker won't punish you for skipping it.

Where do I publish the generated file?

Save it as llms.txt in your site root so it serves at https://yourdomain.com/llms.txt. On most static hosts that is the public/ or root output folder; on WordPress it goes into the uploads root or via a redirect from your root URL.

What pages should I list inside it?

Your homepage, documentation or how-to pages, pricing, and the two or three comparison posts that answer actual buying questions - the pages you would hand a new customer to explain yourself in five minutes.

Do I need anything else besides llms.txt?

No signup and no fee. But llms.txt alone is not visibility: keep your robots.txt open to the AI crawlers, mark up your visible FAQ, and track what the engines actually say about you over time.