Create an llms.txt file for any website in about fifteen seconds. Enter your domain and we scan your sitemap and page metadata, then write a structured file that tells ChatGPT, Claude, Gemini and Perplexity what your business does and which pages matter most.
Completely free. No account, no email, no watermark. Copy the file or download it and publish it at yourdomain.com/llms.txt.
An llms.txt file is a plain-text Markdown file placed at the root of your website that gives large language models a structured overview of your business and your most important pages. It was proposed by Jeremy Howard in 2024 as a way to help AI systems orient themselves on a site quickly, without having to crawl and parse every page. The file starts with your company name as an H1, a one or two sentence summary as a blockquote, and then sections of curated links, each with a short description.
Yes. Enter a URL, get a complete llms.txt file, copy or download it, and use it however you like. There is no account, no email required and no watermark on the output. We ask for an email only if you want to join the waitlist for the full product, and that is entirely optional.
It reads your robots.txt to find your declared sitemaps, falls back to /sitemap.xml, and falls back again to the links in your homepage navigation if neither exists. It then ranks the discovered URLs by page type and depth, fetches the title, meta description and first H1 of the highest-value pages, and passes that structured data to Claude to write the file. Every URL in the output is verified against the list we actually found, so the generator cannot invent a page that does not exist.
They do different jobs. robots.txt controls what a crawler is allowed to fetch. sitemap.xml lists every URL you want indexed, with no explanation of what any of it means. llms.txt is the only one of the three that says what your business is and which pages matter most. It complements the other two rather than replacing either.
Honestly: it is not proven, and anyone telling you otherwise is guessing. Adoption of the standard by major AI providers is still limited, and no provider has published a commitment to reading it. What is true is that the file costs nothing to publish, cannot hurt you, and the exercise of writing one forces useful clarity about your own site. Treat it as cheap insurance on an emerging standard, not as a ranking lever. Strong content and solid technical SEO still do the heavy lifting.
In your website root, the same directory as robots.txt, so it resolves at yourdomain.com/llms.txt. On Next.js that is the public/ folder. On WordPress it is the folder containing wp-config.php. On Shopify you create a llms.txt.liquid template. Webflow and Squarespace do not allow root-level files, so those need a Cloudflare Worker or a proxy in front of the site.
We keep a library of every file the free generator has produced, keyed on domain. When a domain comes back, we serve the stored file instead of scanning the site again. That keeps the tool free and fast for everyone. If your site has changed and you need a fresh file, that is what the full product does — join the waitlist and you can regenerate on demand.
It ranks everything it discovers and then reads the top pages, capped at roughly twenty links in the finished file. Most small business sites are covered comfortably. Large sites will see only their most important pages, which is arguably correct for this format — but if you want a file built from a full crawl plus your company positioning, that is the paid version.
Partially. We read the HTML your server returns, without executing JavaScript. Sites that render their entire navigation and metadata client-side will produce a thinner file, and if such a site also publishes no sitemap we may not be able to generate anything at all. Server-rendered and statically generated sites give the best results.
llms.txt is the index: a curated map of your site with one line per page. llms-full.txt is the proposed companion that inlines the full text of that content into a single file, so a model can ingest everything in one request. This generator produces llms.txt. Start there — the index is the part that is useful even if nothing else on your site changes.
No. llms.txt is descriptive, not permissive. It grants no rights and changes no access controls. If you want to restrict AI crawlers, that is a robots.txt decision — you can block GPTBot, ClaudeBot, Google-Extended and others there. Publishing an llms.txt on a site that blocks AI crawlers is contradictory, so decide the access question first.
Whenever the shape of your site changes: a new product line, a restructured navigation, a retired section, a rebrand. For most companies that is a few times a year. If nothing structural has changed, the file does not need touching.