llms.txt Generator — Build llms.txt from a Sitemap — Free
Turn a sitemap.xml or a list of URLs into an llms.txt skeleton, ready to describe your site to AI assistants.
Frequently Asked Questions
What is llms.txt?
A plain-Markdown file published at the root of a domain that tells AI assistants what a site contains and which pages matter. It is a curated map with descriptions, not a crawl directive — it does not grant or deny permission the way robots.txt does.
Why does it start from a sitemap?
Your sitemap already lists the canonical URLs, so it is the fastest honest starting point. What a sitemap cannot supply is descriptions, which is exactly the part that makes llms.txt worth having — you write those yourself.
Does the tool crawl my site?
No. Browsers cannot fetch arbitrary third-party pages because of cross-origin restrictions, and crawling from your machine would be slow and unreliable anyway. Paste the sitemap you already publish.
Where do I put the finished file?
At the root of your domain, served as text/plain — for example https://example.com/llms.txt. Some sites also publish an expanded llms-full.txt containing the page content itself.
Do AI assistants actually read it?
Support is real but uneven, and the spec is still community-managed rather than formally standardised. IDE agents pointed at documentation sites are the most consistent consumers today. It is cheap to publish and low risk, not a guaranteed traffic source.
Privacy First: All processing happens directly in your browser. Your data never leaves your device.
About llms.txt Generator
llms.txt is a Markdown file at the root of a domain that tells AI assistants what a site contains and which pages are worth reading. This builds the skeleton from a sitemap you already publish; the descriptions are yours to write.
What the file is, and what it is not
The format is deliberately small: an H1 with the site name, an optional blockquote summary, then H2 sections containing link lists where each entry may carry a short description after a colon. That is the whole specification.
It is worth being clear about what llms.txt does not do. It is not robots.txt — it grants no permissions and blocks no crawlers, and publishing one neither invites nor prevents training on your content. It is a curation hint, and an assistant is free to ignore it entirely.
# example.com
> API documentation for the Example platform.
## Docs
- [Getting started](https://example.com/start): install, auth, first request.
- [API reference](https://example.com/api): every endpoint with parameters.
## Notes
- [Changelog](https://example.com/changelog): breaking changes by version.The descriptions are the whole point
A sitemap already gives a machine every URL you have. What it cannot give is judgement about which pages matter and what each one covers. An llms.txt that is only a link list adds nothing that sitemap.xml did not already provide.
This is also the most common failure of auto-generated files: they list everything and describe nothing, which is precisely backwards. A short, curated file with one honest sentence per entry is more useful than an exhaustive one without descriptions.
Deciding whether it is worth publishing
Adoption is real but partial. As of mid-2026 a single-digit percentage of the largest sites publish one, support across assistants is uneven, and the spec remains community-managed rather than formally standardised. IDE agents pointed at documentation are the most consistent consumers today.
The honest assessment is that it is cheap and low risk rather than transformative. Documentation sites, API references and tool directories get the most from it, because those are what assistants are most often asked to consult. A file takes minutes to write and costs nothing to serve, so the calculation is usually favourable — just not on the strength of guaranteed traffic.
Publishing it
- Serve it at the domain root as /llms.txt, with a text/plain content type.
- Use absolute URLs throughout; a model reading the file has no base to resolve relative links against.
- Prune aggressively. Pagination, tag archives and legal boilerplate add noise and crowd out the pages that matter.
- Some sites additionally publish llms-full.txt, which inlines the page content rather than linking to it. That is a much larger file and only makes sense for compact documentation.
- Keep it current. A stale map is worse than none, because it confidently points at pages that no longer exist.