Generate your site's llms.txt.
llms.txt is a Markdown index at the root of the domain: it tells the model what the site is and which pages are worth reading. The tool reads your sitemap and builds the file, ready to publish.
- Module
- AEO/GEO
- Input
- A domain
- Runs on
- Rovemark server
- Cost
- Free, no signup
An index written for a reader that never opens a browser
When an assistant needs to understand your site, it starts with raw HTML: menu, banner, footer, script. The content is there, mixed with everything else. llms.txt is the alternative, a Markdown file at the root, with the site name, one line explaining what it is and the list of pages that matter, grouped by section.
The proposal dates from September 2024, by Jeremy Howard, and has been adopted mostly by documentation and product sites. It is not an official standard and no model is obliged to read it. The cost of publishing is one static file; the gain is no longer leaving the decision of what matters to the parser.
- 01
The tool finds your sitemap
It reads robots.txt looking for the Sitemap directive and, failing that, tries the standard paths. A sitemap index is followed down to its sub-files.
- 02
Pages are grouped by section
The first URL segment becomes a section, it is the hierarchy the site itself has already declared. The shallowest ones come first, and each section gets an allocation, so none consumes the entire file.
- 03
You copy it and publish at the root
The file comes out ready in Markdown. Publish it as /llms.txt at the root of the domain, the same way as robots.txt.
- 01
What is llms.txt?
- A Markdown file at /llms.txt describing the site for language models: title, a one-line summary and links to the main pages grouped by section. The idea is to hand over the map instead of letting the model infer it from HTML.
- 02
Is it the same as sitemap.xml?
- No. A sitemap lists URLs for the search engine crawler, machine talking to machine, without context. llms.txt is written to be read: it has a name, description and grouping, and fits in a context window. One does not replace the other.
- 03
Do models actually read this file?
- There is no guarantee, and it is honest to say so: no major provider has confirmed it consumes llms.txt in production. What exists is growing adoption on the site side and a publishing cost close to zero. It is a cheap bet, not a certainty.
- 04
Where do I publish the file?
- At the root of the domain, as /llms.txt, served as text/plain or text/markdown. If your site is static, drop it in the public folder. Some projects also publish an /llms-full.txt with the whole content, not just the links.
- 05
My site has several languages. Does the file get huge?
- The tool spreads the quota across sections, so each language shows up with a similar number of pages instead of one consuming the whole file. If you prefer one file per language, generate it and cut out the section you need.
- 06
Do I need to update it when I publish a new page?
- Yes, the same as a sitemap. The file is a snapshot of the site at the moment it was generated. If you publish often, generate it at build time rather than by hand.
You published the file. Did the answer change?
llms.txt improves what the model reads. Measuring what it says about your brand, across several assistants and over time, is different work, and that is what the platform does.