llms.txt generator

Summary#

The Free llms.txt Generator fetches your homepage and your sitemap.xml, then builds a ready-to-publish llms.txt file with your site title, description and up to twenty-five of your real pages already listed. Copy it, tidy the labels and upload it to your site root. It runs at /free-tools/llms-txt-generator with no account.

Purpose#

llms.txt is a proposed convention: a short markdown file at the root of your domain that gives AI crawlers and assistants a plain summary of what your site is and a curated list of the pages that matter most. It sits beside robots.txt and sitemap.xml and does a different job. robots.txt sets permissions, sitemap.xml lists everything for indexing, and llms.txt offers meaning: a short description in your own words plus a shortlist of the pages you would want an assistant to rely on.

Writing one from a blank file is tedious. This tool gives you a populated draft in one click.

Overview#

Three fetches run in parallel: the URL you entered, /sitemap.xml at the same origin, and /llms.txt at the same origin (so the tool can tell you whether one is already published). From the page it takes the <title> and the meta description.

What it does with the sitemap depends on the document's root element, not on how the URLs inside it are spelled:

• A <urlset> is a list of pages, and is read directly.
• A <sitemapindex> is a list of other sitemaps. Up to five of its children are fetched and their pages collected. .gz children are skipped, as are children that declare more than 5 MB, and nesting is followed one level only.
• Anything else — most often an HTML error page served with a 200 — is not treated as a sitemap at all.

Up to twenty-five URLs are kept, deduplicated, and restricted to your own domain and its subdomains: a sitemap index may name any host, and the file is published at your root and speaks for your site. Your homepage is always listed first when anything was read.

It then assembles a markdown file and returns it with a copy button and the exact path where it should be published.

Be clear-eyed about what publishing one achieves. llms.txt is a convention, not a standard that every engine follows. Some tools look for it, others do not, and none are obliged to. The honest case is that it costs almost nothing and sits quietly at your root, with more upside as adoption grows. It is not a switch that turns on AI traffic and it is not a Google ranking factor.

Benefits#

• Produces a populated draft rather than an empty template, using your real title, description and pages.
• Derives a readable label for each page from its URL path, so the list is usable immediately.
• Tells you the exact URL the file must be served from.
• One-click copy, no account, no credits.

Use Cases#

• First llms.txt. You have read about the convention and want a starting point in under a minute.
• Refreshing a stale file. Re-generate after a site restructure and diff against what you published.
• Agency deliverable. Generate one per client site as part of an AI-readiness package.
• Closing a check. AI visibility checker, GEO audit and AI citation audit all award points for an llms.txt at your root.

Requirements#

A public domain that returns HTML. A sitemap.xml is optional: without one the tool still produces a file, listing only your homepage.

Permissions#

Anyone. No account, plan or role. A shared hourly fair-use limit applies per IP address across the free tools.

Cost#

Free. Zero credits.

Navigation Path#

metricvaultai.com → Free tools → llms.txt Generator

The page is at /free-tools/llms-txt-generator.

Inputs#

Helper line: No signup · No credit card · Reads your homepage and sitemap.xml.

Step-by-Step Guide#

1. Open /free-tools/llms-txt-generator.
2. Enter your domain in the field marked example.com.
3. Select Generate llms.txt. The button reads Building… and the result area shows Reading your homepage and sitemap.xml….
4. Select Copy. The button confirms with Copied for a moment, then returns to Copy.
5. Paste into a file named llms.txt and upload it to your site root so it is served at https://yourdomain.com/llms.txt. Nothing in it needs rewriting first: every line carries the page's own title and its own description.
6. Load that URL in a browser to confirm it returns your file as plain text.

Reading the Results#

Your llms.txt. A card headed Your llms.txt with a source line beneath it. That line reads one of three things, so you always know where the page list came from:

• From sitemap.xml · <N> pages found — a sitemap was read and gave N pages.
• Sitemap found, but no pages could be read · homepage only — the file was a real sitemap, but nothing usable came out of it (for example, an index whose children were all unreachable).
• No sitemap.xml found · homepage only — there was no sitemap at that URL, or what was served was not one.

The file itself. The generated markdown has a fixed shape:

# <your page title>

> <your meta description, if the page has one>

## Product

- [Example — Everything you need](https://example.com/): The homepage, and what the company does.
- [Pricing](https://example.com/pricing): What each plan costs and what is in it.

## Docs

- [Getting started](https://example.com/docs/getting-started): Install it and run your first report.

## Blog

- [Why we built this](https://example.com/blog/why-we-built-this): The problem that started the company.

## Company

- [About us](https://example.com/about): Who we are and where we came from.

Every line is - [Title](URL): description, which is the shape the llms.txt convention uses.

The title is the page's own <title>, not a label built from the URL. Where a page's title ends with the site name after a dash or a pipe, that tail is trimmed, so a section reads as a list of pages rather than the same brand repeated.

The description is the page's own meta description, falling back to its Open Graph or Twitter description, trimmed to one line. Nothing is written for you: a sentence you have to check is a sentence you have to edit. A page that offers none of the three is left out of the file rather than published as a bare URL, and the result tells you how many were left out.

Pages are grouped into Product, Docs, Blog and Company by their path, in that order, and a heading only appears when it has something under it.

Sign-in, app, cart and account paths are never listed. /login, /signup, /account, /cart, /checkout, /dashboard, /admin, /wp-admin and the rest of that vocabulary are dropped before the page list is built, so they never take a slot a real page could have had. These pages are not secret — they are simply no use to a model, and a cart is actively unhelpful to advertise. Real sitemaps do list them: huel.com's names /login, /cart, /register and /recover.

Where to put it. A note below the file repeats the destination:

Examples#

Screenshots#

Tips#

• Curate before you publish. A short, honest file beats a dump of every URL, and the generated list is a starting point rather than a finished one.
• Rewrite the labels. Slug-derived names such as getting started v2 read badly in an answer.
• Read the descriptions once. They are your own meta descriptions, so a page with a weak one produces a weak line - and that is worth knowing either way.
• Re-generate whenever you launch a product, change pricing or restructure your docs. A stale file points assistants at dead ends with full confidence.

Best Practices#

• Keep the summary line accurate and specific: what you sell and who it is for. That sentence is the one an assistant is most likely to repeat.
• Link the pages you can afford to be quoted on: core product, pricing, docs, key policies and contact.
• Keep robots.txt and sitemap.xml as they are. llms.txt does not replace either, and removing them to "simplify" would be a serious mistake.
• Put the file in version control alongside the rest of your site so it is reviewed like any other page.

Common Mistakes#

• Assuming it is complete. The file lists up to 25 pages from your sitemap, which on a large site is a fraction of it. Add the pages that matter most to you and drop anything you would rather an AI engine did not summarise.
• Uploading it to the wrong path. It must be served from the root, at /llms.txt, not from a subdirectory.
• Expecting a ranking or traffic effect. There is no evidence for one and we do not claim it.
• Letting it go stale. A broken link here sends an assistant somewhere wrong and confident.

Limitations#

• Twenty-five pages maximum, taken in the order they appear in your sitemap. There is no relevance ranking, so the pages you care about may not be among them.
• Only /sitemap.xml is looked for. A sitemap published anywhere else, or declared only by a Sitemap: line in robots.txt, is not discovered.
• A sitemap index is followed one level deep, and only its first five children are fetched. A site that splits its pages across more than five child sitemaps will be sampled, not read in full.
• Compressed (.gz) child sitemaps are skipped.
• URLs on a different domain are dropped, even when your sitemap index lists them. Subdomains of your own site are kept.
• The title and description come from whatever the homepage serves. If the page has no meta description, the > summary line is omitted entirely.
• The tool does not publish the file for you and cannot check that you did.
• There is no in-app version of this generator. The dashboard's technical tools cover robots.txt and sitemaps, not llms.txt authoring.
• No export, history or scheduling. See Free tools, no account needed.

Troubleshooting#

FAQs#

What is llms.txt? A plain markdown file published at the root of your site that gives AI crawlers a short summary of what you are and a curated list of your most important pages.

How is it different from robots.txt and sitemap.xml? robots.txt sets crawl permissions. sitemap.xml lists every URL for indexing. llms.txt does neither: it hands an assistant a plain-language summary and a shortlist. Keep all three.

Do AI engines actually read it yet? It varies, and we will not overstate it. It is a proposed convention rather than an official standard. Some tools look for it, others do not, and none are obligated to.

Does it help my SEO? It is not a Google ranking factor and we would not promise a traffic change. Its value is in helping assistants describe and cite you from a clean source.

How often should I update it? Whenever your pages, product or pricing change materially, and a quick read every quarter.

What does the paid product add? The generator builds the file. The app tracks whether the answer engines actually mention and cite your pages over time, and how you compare with competitors. See Visibility Overview and What you get on Free.