llms.txt Generator

Turn your site into a guide AI assistants can follow. Scan a URL to draft the file from your navigation and sitemap, edit the page list, then copy or download an llms.txt ready for your domain root.

Scan your site
We read your homepage, navigation and XML sitemap, then draft the file for you. Nothing is stored.
No account, no email, no install.

Crawl & discover

Homepage links plus sitemap URLs, filtered to pages worth citing.

Edit the draft

Rename pages, pick sections, and mark anything skippable as Optional.

Ship the file

Copy or download llms.txt, then publish it at your domain root.

What an llms.txt does

/llms.txt is a Markdown file at your domain root that gives AI tools a curated map of the site: what it is, who it is for, and the handful of pages worth reading. It is the same idea as robots.txt, except it guides instead of restricting.

An assistant working from raw crawl data has to guess which of your pages matter. A good llms.txt answers that in one fetch — the homepage, product pages, docs, pricing and the comparisons you actually want quoted.

It is a convention, not a ranking factor. Googlebot does not need one and no engine is obliged to read it. What it does is make the right answer cheap to find when a model does look.

Best-practice structure

The format is small and strict: an H1 name, a blockquote summary, an optional detail section, then link lists under H2 headings.

# Acme

> Acme is a privacy-first analytics platform for product teams.

Acme tracks pageviews, goals and revenue without cookies or
cross-site profiles. This file points assistants at the pages
that explain the product and how to install it.

## Docs

- [Install guide](https://acme.com/docs/install): Add the script in one line.
- [API reference](https://acme.com/docs/api): Read-only endpoints and auth.

## Pricing

- [Plans](https://acme.com/pricing): Free trial, then usage-based plans.

## Optional

- [Changelog](https://acme.com/changelog): Shipped changes, newest first.

Tips that matter

  1. 1Keep it short. Aim for 20–60 links. An llms.txt is a shortcut past the crawl, not a sitemap replacement — the full list already exists in sitemap.xml.
  2. 2Write descriptions that answer a question. Each link takes an optional description after a colon. Say what the page covers or what question it answers, in one clause.
  3. 3Use absolute URLs. Assistants read the file out of context. Relative paths and tracking parameters make the link harder to follow and duplicate.
  4. 4Put the pages you want cited first. Sections are read top to bottom. Lead with the overview and product pages, follow with docs and pricing, then blog and legal.
  5. 5Use Optional for skippable links. The spec treats an "Optional" section as secondary. Changelogs, legal pages and archives belong there, so assistants can skip them.
  6. 6Serve it as text/plain at the root. The file must answer at https://yourdomain.com/llms.txt with a 200 and no redirect chain. A styled HTML page or an auth wall will not be read.
  7. 7Keep robots.txt and llms.txt consistent. A file that invites assistants to pages robots.txt blocks is worse than none. Check crawler access first.
  8. 8Regenerate when the site changes. A stale llms.txt sends assistants to moved docs and retired pages. Re-run the generator after a relaunch or a docs migration.

llms.txt vs llms-full.txt

llms.txt is the index: short, curated, cheap to read. llms-full.txt is an optional companion holding the full text of the important pages in one file, for tools that want depth in a single request.

Publish the index first. Add the full variant only if you have documentation that is genuinely worth pasting end to end — it goes stale quickly and can grow into megabytes.

Next steps

Publish the file, then confirm assistants can reach it: the AI crawler access checker verifies robots.txt rules for every major AI bot. The AI readiness scorecard covers the rest of the technical picture — schema, semantic HTML and discovery files.