llms.txt generator: a starter file from your sitemap
Paste a web address. We read the home page, robots.txt and the sitemap the way a crawler does, open up to twenty listed pages for their titles, and write a starter llms.txt in the convention's shape. It is a draft, marked as one: keep the pages that matter, add a line to each, publish it at /llms.txt.
Requests are sent as QuietwatchBot, never as another bot; robots.txt is obeyed for every page we open. Free: 5 runs an hour per address.
What is llms.txt?
A plain-text Markdown file at the root of a site, proposed in 2024 by Jeremy Howard of Answer.AI, that gives language models and their crawlers a short, curated index of the site: a title, a one-line summary in a blockquote, then sections of links with one-line notes. It is a convention, not a standard anyone enforces; the vendors' crawler pages do not say they read it. quietwatch checks for it on every site because a readable index is cheap to publish and some assistants and tools do fetch it.
What does the generator write?
| Part | Where it comes from | What to do with it |
|---|---|---|
| The H1 | The home page's <title> | Shorten it to the site's name |
| The summary (blockquote) | The home page's meta description, or a placeholder | Write one honest line: what the site offers and for whom |
| Pages | Up to twenty URLs from the sitemap, same host, each opened for its title and description | Delete what does not matter; keep pricing, docs, contact and the pages people cite |
| Optional | An empty section | Add policies, changelog, anything secondary |
How to publish it
- Download the draft, edit it in any text editor, and keep it under a few hundred lines.
- Upload it to the site root so it answers at
/llms.txtwith the content typetext/plain, not an HTML page (a catch-all route is the usual mistake). - Run the free checker: the llms.txt row should say present, with the byte count and a content hash. quietwatch then watches it daily for every site you add.
Questions
Will this make AI assistants cite my site?
No one can promise that. llms.txt is a convention: it gives a crawler a short, curated index. Whether a given assistant reads it is not documented by the vendors. What is certain is that it costs nothing, and quietwatch can watch it for you so a deploy never deletes it unnoticed.
Why is it called a draft?
Because the generator only knows your sitemap and your page titles. You know which pages matter and how to describe them in one line. Ten good links beat two hundred automatic ones.
Which pages does it include?
Up to twenty URLs from your sitemap, on the same host, in sitemap order, skipping files such as PDFs and images, and skipping any page robots.txt keeps our crawler away from. The limit keeps the draft readable and the fetch polite.
Where should it live?
At the site root: https://your-site.example/llms.txt, served as text/plain. The llms.txt check on the free checker confirms it the moment it is up.
Does the generator read the site in any special way?
No. It fetches the home page, robots.txt, the sitemap and the listed pages with our own User-Agent, exactly as the free checker does, and runs no JavaScript. Nothing is stored except the draft, which the link shows for 30 days.