S SCHEMA.BIZ

Last updated:

llms.txt Generator

This llms.txt generator drafts the file from your site's sitemap and pages, for you to edit.

Our server fetches the home page, the sitemap (/sitemap.xml, or the one robots.txt lists) and up to 20 pages as SchemaBizBot; the draft is built and edited here in your browser.

How to use the llms.txt generator

Enter your site's address. Our server fetches its home page as SchemaBizBot, for the site's name (from og:site_name or the title) and the meta description that becomes the summary. It then reads /sitemap.xml (or, when there's none, the first sitemap the site's robots.txt lists), or the first sitemap of a sitemap index, takes up to 200 of its addresses and groups them by their first path segment: /blog/… becomes a "Blog" section, /shoes/… a "Shoes" section, and pages directly under the root join the home page in "Pages". Finally it fetches up to 20 of those pages, 4 at a time, spread across the groups, and uses each one's title (without the site's name) as the link name and its meta description as the notes.

The draft appears beside the file it makes. Rename the site and sections, rewrite the summary, edit, remove or add links, and add sections; name one "Optional" for links an agent may skip. The file updates as you type and is checked with the same parser as the llms.txt validator. When it reads well, copy it or download llms.txt. "Start from scratch" opens an empty draft without fetching anything. A link such as /ai/llms-txt-generator/?url=https://example.com/ starts the draft when it opens.

A page that doesn't load, or that our server couldn't fetch just then because of too many checks in a short time, is left out, and the summary above the draft says how many. Run it again a minute later to fill the gaps, or add those links by hand.

What a good llms.txt looks like

A short, curated map: the name, one or two sentences about the site, and the pages that answer the questions people ask about you, each with a note that says what's there. For a small shop:

# Example Shop

> Trail running shoes and gear, sold online and shipped from Denver, Colorado.

## Pages

- [Trail running shoes and gear](https://www.example-shop.com/): The shop's home page
- [About us](https://www.example-shop.com/about/): Who runs the shop and where we ship

## Shoes

- [Trail shoes](https://www.example-shop.com/shoes/trail/): Every trail model, with sizes and weights
- [Road shoes](https://www.example-shop.com/shoes/road/): Cushioned and racing road shoes

## Optional

- [Blog](https://www.example-shop.com/blog/): Running guides and product news
  • Choose, don't dump. It isn't a sitemap: a few dozen well-described links say more than every address you have.
  • Write the notes for a reader who hasn't seen the page. "Every trail model, with sizes and weights" says more than "Trail shoes".
  • Keep the summary factual. What the site is, who it's for, where it operates; not a slogan.
  • Use full URLs, and put background reading under "Optional".

The format comes from the proposal at llmstxt.org: an H1 with the name, a blockquote summary, optional free text without headings, then H2 sections of [name](url): notes links. Our guide to llms.txt explains where it came from and what it can and can't do.

Before you publish

An llms.txt points at pages; it doesn't let anyone in. Check that AI crawlers may fetch those pages with the robots.txt checker, and that their structured data says what they are with the AI Search Optimizer. After uploading, run the validator on your live address: a host that answers every unknown address with its home page will serve HTML at /llms.txt until the file is really there.

Frequently asked questions

FAQ

Will an llms.txt get my site cited by ChatGPT or Google's AI features?

Nobody can promise that. llms.txt is a proposal, and support from the major AI providers is limited or unconfirmed: we haven't found one that documents reading it. It's quick to make with this generator, so it's a reasonable optional step, but the documented parts come first: let AI crawlers in, and give pages clear text and valid structured data.

Which pages does the generator pick?

Up to 20 pages from the first 200 addresses in the site's sitemap, taken in turn from each group of pages that share a first path segment, shortest addresses first, so a section's index page comes before its articles. The home page is always included. Pages that don't load are left out. Edit the result: add the pages that answer the questions people ask about you, and remove the rest.

What if my site has no sitemap?

The draft starts from the home page alone, and you add sections and links by hand. The generator looks for /sitemap.xml; when there's none, it reads the first sitemap your robots.txt lists on a Sitemap: line, and it says which one it used. For a sitemap index, it reads the index's first sitemap. The sitemap checker shows whether yours is where crawlers look.

Where do I put the file?

Upload it to your site's root, so it answers at https://example.com/llms.txt, served as text/plain or text/markdown. On most hosts that means placing it in the folder that holds your robots.txt; in a static site generator, its public or static folder.

Does the generator store my site's pages?

No. Our server fetches the home page, the sitemap and the pages as SchemaBizBot and sends them to your browser, where the draft is built. What you type into the draft stays in your browser. Like every tool on the site, the generator records what kind of run it was: whether a sitemap was found, how many pages were read or failed, and how many sections and links a copied file has, never the pages, their titles or your text. The site's address is recorded without its query string. See the privacy policy.