Check your llms.txt,
or write one.
Type a website. The validator reads its live /llms.txt, runs eight checks based on the llms.txt proposal and shows what an AI agent would get. No file yet? The generator below writes one for you to copy.
The eight checks
The llms.txt proposal keeps the format small on purpose: a Markdown file at the root of a site that tells AI agents what the site is and which pages matter. The validator checks the parts an agent relies on.
- Served as plain text. Agents read Markdown. An HTML content type usually means a site template has caught the address.
- A # title and a > summary. The title names the site and the one-paragraph summary tells an agent what it is before it decides what to read.
- ## sections with at least three links. The value is in the choice of pages. A list of every address you own is a worse sitemap, not a better guide.
- Full URLs. The file is often read away from your site, where a relative link such as
/pricingpoints nowhere. - No HTML and under 100 KB. Markup suggests the wrong thing is being served, and a file that large has stopped being a short guide.
The validator also notes whether the optional llms-full.txt exists, since agents that read one often look for the other.
The failures you can't see in a browser
The commonest failure is a file that isn't one. Many sites answer any unknown address with their home page or a styled not-found page, so /llms.txt loads something in a browser while every agent that asks for the file gets HTML. The next is a file built from relative links, which reads fine on your own domain and breaks everywhere else. The third is a file behind a login, a bot wall or a robots.txt rule that blocks the agents meant to read it: check that with the free AI crawler checker.
Write an llms.txt file
Fill in the fields and the file builds as you type. List one page per line as /path | Label | short note, and start a line with ## to begin a new section. A section called ## Optional marks pages an agent may skip when it is short of space.
Before you publish it
Save the file as llms.txt at the root of your site, so it loads at https://yourcompany.com/llms.txt as plain text for anyone who asks, without a login. Then run the validator above on your site to confirm it reads the way you meant.
Keep it current. A short guide that points agents confidently at pages that have moved is worse than none, so review it whenever your main navigation changes. Our guide to what llms.txt is and whether you need one covers the proposal in more detail, and IndexHalo's own file is a working example.
What a pass does and doesn't mean
A pass means an agent that reads the file will understand it. It doesn't mean the file lifts your rankings: Google says you don't need new AI text files to appear in AI Overviews or AI Mode. Crawler access and clear pages matter more. This check says who may read a site. It can't say what ChatGPT, Gemini, Claude, Perplexity or Google's AI answers say when a buyer asks about your category: only a search can. IndexHalo runs those searches and keeps every answer and source, on plans from $49 a month for 100 searches.
Questions people ask
What does the llms.txt validator check?+
It fetches the live /llms.txt from the site's root and runs eight checks based on the llms.txt proposal: that the file is served as plain text, opens with a # title, has a > summary and ## sections, links at least three pages with full URLs, contains no HTML and stays under 100 KB. It also notes whether llms-full.txt exists.
Why does the validator say my llms.txt address returns a web page?+
The site answered /llms.txt with an HTML page, usually because a catch-all route serves the home page or a not-found page for any address it doesn't know. An agent asking for the file gets that page instead. Publish a real text file at that address and serve it as text/plain.
What is the difference between llms.txt and llms-full.txt?+
llms.txt is a short guide: a title, a summary and lists of links to the pages that matter. llms-full.txt is an optional companion that holds the full text or much longer descriptions in one file, for agents that want the content without fetching every page. If you publish both, keep them in step.
Does a valid llms.txt improve AI visibility?+
Not in any way you should plan around. Google says you don't need new AI text files to appear in AI Overviews or AI Mode, and no AI company has said it ranks pages by llms.txt. A valid file helps the agents that do read one, which matters most for documentation sites.
Should the links in llms.txt be full URLs?+
Yes. An agent often reads the file away from your site, and a relative link such as /pricing has nothing to resolve against there. The generator on this page turns every path into a full address on your domain, so its output passes that check.