Generate an llms.txt file from your sitemap
Give it your domain and it reads your sitemap, groups your pages into sections and writes the file. Then the part most generators leave out: no major AI engine has said it reads llms.txt yet. It costs twenty minutes and it may pay off later. Four other things pay off now, and they are listed below the tool.
No account, no card. We read your sitemap and nothing else.
What llms.txt is
A markdown file at the root of your site. It names your important pages and gives each one a line of context, so a model reading your site gets a curated map instead of a sitemap full of bare URLs. The format borrows from robots.txt in placement and from a table of contents in spirit.
What it looks like
# Company Name
> One line on what you do and who it is for.
## Docs
- [Getting started](https://site.com/docs/start): what a new user does first
- [API reference](https://site.com/docs/api): endpoints and authentication
## Product
- [Pricing](https://site.com/pricing): plans and what each includesWhether it does anything
Nothing measurable, yet. No major engine has said it reads the file: not OpenAI, not Anthropic, not Google, not Perplexity. Some smaller tools do. Anyone telling you it is required is selling you something, and the thing they are selling is usually a subscription.
Publish one anyway if you like. It takes twenty minutes, writing it forces you to say what each page is for, and half the sites we look at cannot answer that question. If support arrives you are already there.
The four things that do work right now
1. Make sure the crawlers can reach you
This is the one that quietly ruins everything else. A CDN rule can return 403 to every AI crawler while your robots.txt says come in, and nothing in your analytics reports it. Check yours, it takes ten seconds.
2. Answer the question in plain text, near the top
Models lift passages, not pages. A self-contained answer in the first hundred words gets quoted; the same answer buried under a brand story does not. Real HTML text, not an image of text, and not a table built out of flexbox divs.
3. Get mentioned where the engines already read
Being named in an independent roundup beats a page on your own domain calling yourself the best, because the engine trusts a third party more than a vendor. Run your target question in Perplexity and read the citations: that list is your target list.
4. Keep your facts consistent
Three different prices on three pages teaches a model to trust none of them. Same for your founding year, your headcount and what you claim to do.
llms.txt is roughly ninth on that list. Do the first four.
Questions
What is llms.txt?
A markdown file at the root of your site that lists your important pages with a line of context each. The idea is that a model reading your site gets a curated map instead of guessing from a sitemap full of bare URLs.
Do the AI engines actually read it?
As of today no major engine has said it does. Not OpenAI, not Anthropic, not Google, not Perplexity. Some smaller tools read it. Anyone telling you it is required is selling something.
Then why publish one?
Three reasons, none of them thrilling. It costs twenty minutes. Writing it forces you to say what each page is for, and plenty of sites cannot answer that. And if support arrives you are already there. If it never arrives you lost twenty minutes.
What should I do instead, or first?
Make sure the AI crawlers are not blocked, which is the one that quietly ruins everything else. Answer real questions in plain text near the top of your pages, because models lift passages rather than pages. Get mentioned on the pages the engines already read. Keep your facts consistent across your site. All four beat llms.txt.
Where do I put the file?
At the root, served as plain text: https://yoursite.com/llms.txt. Nothing else needs to change.