Back to Glossary

Glossary

llms-full.txt: What It Is and How to Use It

Updated September 2026TechnicalIntermediate

Definition

llms-full.txt is a single plain text file holding the full content of a site’s key pages, not just links to them. llms.txt is the shorter companion, a curated index pointing at where things live. Both are proposed conventions, and no major AI lab has confirmed that it reads either one.

At a glance

  • llms-full.txt sits at yoursite.com/llms-full.txt and is written as plain Markdown.
  • The file carries the text of your pages rather than links, which makes it large.
  • A documentation set or a glossary is the realistic candidate. A blog archive is not.
  • Adoption is unconfirmed, so treat the file as a cheap experiment rather than a channel.

One file, your whole reference section

The idea is to remove the round trips. Rather than fetching thirty documentation pages one at a time, a consumer reads the set in a single pass.

The llms.txt proposal covers both files. The short one is an index of links with one-line descriptions. The long one is the content itself, concatenated as Markdown with headings preserved.

There is no schema and no size rule, which is the weak point. A full-text dump of a large site runs to megabytes, and nothing in the convention says what anyone should do with that.

Keep it to a section that reads as a unit: documentation, a glossary, an API reference.

Where a full-text file earns its keep

Everyone wants this to be a distribution channel. It is closer to a convenience.

The case for it is narrow and real. If you publish a reference set an assistant would want whole, one clean Markdown file removes every chance of it hitting a broken render, a cookie wall or a page skeleton waiting on server-side rendering that never arrives.

The case against is that nothing is obliged to read it, and no lab has said it does. A file nobody has committed to fetching is not a plan.

So size the project accordingly. It is a generated artifact, not a content program.

Generate it from what you already publish

  1. Pick one section that reads as a set. If you cannot name it in three words, skip the whole thing.
  2. Export those pages to Markdown from your CMS, headings intact, navigation and footers stripped.
  3. Concatenate into one file with a heading per page and the source URL underneath each, so attribution survives.
  4. Serve it at /llms-full.txt as plain text. Confirm it does not come back wrapped in your site template.
  5. Regenerate it in your build. A hand-maintained copy goes stale within a month, and a stale file is worse than none.

Leave this alone until crawlability is sorted. A crawler that cannot read your pages is not rescued by a file repeating them.

llms.txt vs llms-full.txt

  llms.txt llms-full.txt
What it holds Links with one-line descriptions The full text of those pages
Typical size A page or two Whatever your reference section weighs
Effort About an hour by hand Only sensible if generated
How fast it goes stale Slowly Every time you edit a page
Worth it when You want to signal priority You publish a reference set

Frequently asked questions

  • Do AI crawlers actually read llms-full.txt?

    No lab has confirmed that it does. Some agents fetch the file, which server logs will show, but a fetch is not evidence that anything downstream changed. Publish it because generating it is nearly free, not because anyone has promised it counts.

  • Is llms-full.txt the same as a sitemap?

    No. A sitemap lists URLs so a crawler can discover them, and carries no content. llms-full.txt carries the content and lists nothing for discovery. A sitemap is a long-standing protocol every crawler uses. This is a proposal with unconfirmed uptake.

  • How large should the file be?

    Small enough that reading it beats crawling the pages, which in practice means one section rather than a whole site. If you are trimming a multi-megabyte export to make it usable, this was the wrong tool for that content.

  • Should I write llms.txt or llms-full.txt first?

    llms.txt, and possibly only that. It takes an hour, it is easy to keep current, and it covers the whole site. The full-text version only makes sense once you have a reference section worth handing over whole and a build step that can produce it.