How to use llms.txt generator and validator
- Enter your site's name and a one-line summary.
- Optionally enter your website address and select Draft from sitemap, or start blank.
- Edit the titles and add short notes to the links that matter most.
- Fix anything the validator flags, then download llms.txt and publish it at your site's root.
What goes in an llms.txt
The format, proposed at llmstxt.org, is plain Markdown. It starts with your site's name as a level-one heading, then a one-sentence summary in a blockquote, then optional free text, then sections under level-two headings. Each section is a list of links in the form - [Title](https://…): a short note. A section called Optional holds links a model may skip when it's short of space.
The point is curation. A sitemap lists every page; llms.txt lists the pages that best explain what you do, with notes a model can use to pick the right one. Ten to fifty well-described links beat five hundred bare ones.
Drafting from your sitemap
Draft from sitemap reads the sitemap your robots.txt names, or /sitemap.xml, follows a sitemap index into up to five child sitemaps, and groups the first two hundred pages by the first part of their path, so /docs/… pages form a Docs section and /blog/… pages a Blog section. Titles are made from the page addresses, which is a starting point: rename them, delete pages that don't help, and add a note to each link you keep.
Only the website address goes to our server, which reads the public sitemap and returns the list of page addresses. Editing and validation happen in your browser.
Validation
The validator checks the file as you type: one H1 title at the top, a summary blockquote, sections at level two, list items that are links, full https:// addresses so links work out of context, and no empty sections. Paste an existing llms.txt into the box to check it the same way. When it's clean, download the file and publish it at the root of your site, for example https://example.com/llms.txt.
Example
A small llms.txt might read: # Acme Docs, then > Acme is an invoicing API for marketplaces, then ## Guides with - [Quickstart](https://acme.dev/start): create an account and send the first invoice, and - [Webhooks](https://acme.dev/webhooks): events and retries, then ## Optional with the changelog. That's enough for a model to find the right page for most questions.
Questions
- What is llms.txt?
- A Markdown file at /llms.txt, proposed at llmstxt.org, that gives AI models a short, curated map of a site: its name, a summary, and lists of the most useful pages with notes.
- Does llms.txt replace robots.txt or a sitemap?
- No. robots.txt says what crawlers may read, a sitemap lists every page for search engines, and llms.txt is a short guide for language models. Keep all three.
- What is sent to your server?
- Only the web address, and only when you use Draft from sitemap: our server reads the site's public sitemap and returns its page addresses. Editing and validating happen in your browser.