AI tools like ChatGPT and Claude are starting to look for a curated map of your site’s best pages. With this free tool, I take out the guesswork and format your links into a spec-correct llms.txt file, checked field by field against the real llmstxt.org standard, ready to copy, download, and drop at your domain root.
Build a spec-correct llms.txt file by hand, no crawling involved. llms.txt is a curated file, so you pick exactly which pages and links belong in it, and this tool formats them correctly and lets you copy or download the result.
Becomes the H1, the only section the spec actually requires.
Rendered as a blockquote directly under the H1. Keep it to one line.
Rendered as freeform paragraph(s) after the summary, before your link sections.
Group your links under H2 headings, Docs, Guides, API Reference, whatever makes sense. A section literally named Optional is reserved by the spec: it signals those links can be skipped for a shorter context, so this tool always places it last no matter where you add it.
/llms.txt at your domain root, the same level as robots.txt.No crawling and no waiting on a live URL. You provide the structure, and the generator formats it exactly the way llmstxt.org describes.
Your site name becomes the H1, the only section the spec requires. A one line summary renders directly underneath it as a blockquote.
Add as many H2 sections as you need, Docs, Guides, whatever fits your site, then add link rows with a title, a URL, and an optional note for each one.
Name a section “Optional” and the generator recognizes the reserved keyword automatically, moving it to the end of the file no matter where you added it.
Grab the finished markdown with one click, then upload it as /llms.txt at your domain root, the same level as robots.txt.
This isn’t just a text box. Each field is validated against llmstxt.org’s actual structure in real time, so you catch a missing summary or an incomplete link row before you ever download the file.
Everything else is optional. The generator flags it immediately if your site name is missing, since that’s the only section llmstxt.org actually requires.
A link needs both a title and a URL to make it into the file. Rows with just one filled in are flagged and quietly skipped, not published broken.
Name a section “Optional” and it’s automatically sorted to the very end of your file, exactly as the spec intends, so shorter context tools know what to skip first.
Explicit pass confirmations for your H1, summary, and section formatting, not just a list of problems, so you know exactly what’s ready.
Copy the generated markdown straight to your clipboard, or download it as llms.txt and drop it at your domain root, right where robots.txt already lives.
There’s no evidence of that yet. Ahrefs studied roughly 38,000 domains with a valid llms.txt file and found 97 percent got zero requests for it in a full month, no bots and no humans. Publishing one is low effort and low risk, not a guaranteed result.
Chrome’s Lighthouse tool now includes an agentic browsing audit that looks for llms.txt, but the file is still optional. A clean 404 is graded as not applicable rather than a failure. Google’s own John Mueller has said no AI system currently uses llms.txt, and compared it to the old keywords meta tag, a claim about your own site that a crawler has no obligation to trust.
A sitemap lists every indexable URL with no descriptions and no curation. llms.txt is the opposite: a short, hand picked list of your best pages, with room for notes on each one, and it can even point to pages outside your own domain.
It’s a reserved heading per the spec. Tools reading your file that need a shorter context window can skip anything listed under Optional first. The generator recognizes the name automatically and always renders that section last, no matter where you added it.
Adoption is still small and skewed toward technical, AI aware sites. Semrush found roughly 951 domains had a published llms.txt as of mid 2025. Ahrefs found about 28 percent of the 137,000 domains it studied had one, but that sample leans heavily toward site owners already paying attention to AI search, not the average site.
No. Every field, the live preview, and the download all run in your browser. Nothing is uploaded, logged, or stored anywhere.
The research behind this tool, plus a few more free tools to round out your technical SEO checklist.
→
→
→
→
