# all·markdown

llms.txt generator

Build a spec-shaped llms.txt — an H1, a summary blockquote, then sections of links. Fill the form, or paste a sitemap and let it fill the links for you. Everything happens in this tab; nothing is uploaded and nothing is crawled.

Paste a sitemap or a list of URLs instead

Parsed here in the page. Titles are guessed from the last part of each URL and are meant to be edited.

What the file has to contain

The convention is strict about shape and loose about everything else. In order:

  1. An H1 with the name of the site. The only required line.
  2. A blockquote summarising what it is, in one or two sentences.
  3. Optional prose — anything a reader needs before the links make sense.
  4. H2 sections holding link lists, each entry written - [Title](url): what it covers.
  5. An ## Optional section, last.

That last section is the one most people leave out and the one doing the real work: it tells a model with a small context budget what to drop first. Name a section Optional here and it is always written last, whatever order you added it in.

Where to put it

At the root of the domain — example.com/llms.txt — served as text/plain or text/markdown. Some sites also publish an llms-full.txt with the page content inlined rather than linked, for tools that would rather have one file than follow twenty.

Link to Markdown versions of your pages where you have them. That is the point of the exercise: a page fetched as .md costs a fraction of the same page fetched as HTML, and none of the budget goes on navigation.

What it is not

It is not robots.txt. That file is about permission, is read by crawlers before they fetch, and is enforced by convention. llms.txt is about orientation, is read at inference time, and enforces nothing. Publishing one blocks nothing and grants nothing — you are offering a map.

It is also not a ratified standard. It is a proposal that has been widely adopted, and no major model provider has committed to reading it. The cost of publishing one is a few minutes, so treat it as cheap insurance rather than as something with a guaranteed effect.

Common questions

What is llms.txt?
A Markdown file at the root of a site — example.com/llms.txt — that describes what the site contains and links to the pages worth reading. It exists because a language model given a rendered web page spends most of its context on navigation and markup rather than on content.
Where does the file go?
The root of the domain, served as text/plain or text/markdown. Large sites often also publish llms-full.txt with the whole content inlined, for tools that would rather have one file than follow links.
Does this crawl my site?
No, and it could not — everything here runs in your browser, which cannot fetch other people's pages. Paste a sitemap.xml or a list of URLs and it is parsed in this tab. Nothing you type reaches a server.
What is the Optional section for?
It marks links that can be skipped when context runs short. It is the part most people leave out and the one doing real work — without it a model with a small budget has to guess what to drop. Name a section "Optional" and it is always written last.
Is it the same as robots.txt?
No. robots.txt is about permission and is read by crawlers before they fetch. llms.txt is about orientation and is read at inference time. Nothing is blocked by publishing one — you are offering a map, not setting a rule.
Is llms.txt an official standard?
It is a proposal that has been widely adopted rather than a ratified standard, and no major model provider has committed to reading it. The cost of publishing one is a few minutes; treat it as cheap insurance rather than as something with guaranteed effect.