What does a valid llms.txt file look like?

Markdown with a fixed shape: exactly one H1 title as the first line, an optional one-line blockquote summary right after it, optional free-form context, then H2 sections whose contents are link lists in the form '- [Link text](https://url): short description'. An 'Optional' section at the end marks links agents may skip when they need a shorter context.

Do AI engines actually read llms.txt?

Mostly no, and we won't pretend otherwise. Ahrefs measured that 97% of published llms.txt files get zero requests from AI crawlers, and Google has said it doesn't use the file — John Mueller compared it to the meta keywords tag. Where it does get used is on-demand: coding assistants, agents, and dev tools fetch it when a user points them at your site.

If almost nobody reads it, why validate it?

Because the consumers that do exist fetch it programmatically and parse it strictly. A file with broken links, relative URLs, or an HTML error page served at the path is worse than no file — the agent wastes its context window on garbage or dead ends. If you publish one, it should be small, valid, and truthful.

What's the most common mistake this validator finds?

HTML served at the /llms.txt path. Many sites return their SPA shell or a soft-404 page with HTTP 200 for any unknown path, so tools that fetch llms.txt get a web page instead of markdown. The second most common: sitemap dumps — hundreds of bare URLs with no descriptions, which defeats the point of a curated index.

What's the difference between llms.txt and llms-full.txt?

llms.txt is a curated index: title, summary, and annotated links. llms-full.txt is the companion convention (popularized by Mintlify and Anthropic) that inlines the full text of your pages into one file. This validator checks the index file, and in URL mode also tells you whether an llms-full.txt exists at the same origin.

Free toolsChecker

llms.txt Validator

Check your llms.txt against the llmstxt.org format: structure, link syntax, broken links, and the misconfigurations that make a file useless — with line-by-line results and a score.

We fetch yourdomain.com/llms.txt, validate the structure, and check every linked URL (up to 30) for broken links.

How it works

1
Enter your domain or paste the file
URL mode fetches yourdomain.com/llms.txt directly; paste mode validates content in your browser before you publish it.
2
We check structure, syntax, and links
Every rule from the llmstxt.org format — one H1, blockquote summary, H2 link sections, absolute URLs — plus live broken-link checking in URL mode.
3
Fix what's flagged
Findings are grouped by severity with line numbers, and the annotated source view highlights exactly where each problem is.

The honest state of llms.txt in 2026

llms.txt is a community proposal from Answer.AI (September 2024), not a standard any major AI engine has committed to. The adoption picture is lopsided: roughly 10% of sites publish one (SE Ranking), including Anthropic, Stripe, and Cloudflare — but Ahrefs found that 97% of published llms.txt files receive zero AI-crawler requests, and Google has said flatly that it doesn't use the file. Any validator that implies llms.txt is a ranking factor is selling you something.

So why does this page exist? Because the real consumers of llms.txt aren't crawlers — they're agents and dev tools fetching it on demand: a coding assistant pointed at your docs, an agent researching your product, an IDE integration pulling your API reference. Those fetches are invisible in crawler stats, they parse the file strictly, and they happen exactly when someone is actively trying to use your site. For that audience, a broken llms.txt is worse than none at all.

What the validator checks

Structure: exactly one H1, and it must be the first content element; a blockquote summary immediately after it; no H3+ headings; free-form context allowed only before the first H2. Sections: every H2 section should contain only - [text](url): description list entries — malformed entries, empty link text, and empty URLs are errors. URLs: must be absolute http(s) — relative URLs are dead ends for a model reading the file out of context; plain-http links and duplicates are warnings.

Quality smells: more than half the entries missing descriptions, files over 100 KB, and the classic sitemap dump — a hundred-plus bare URLs pasted in with no curation. In URL mode we additionally fetch every linked URL (up to 30) and flag broken links, detect HTML being served at the llms.txt path, and check whether an llms-full.txt exists at the same origin. Scoring is simple and transparent: start at 100, minus 15 per error and 5 per warning.

Fix it, then measure what actually matters

If your file fails validation, the fastest fix is usually regeneration: the free llms.txt generator builds a spec-correct file from your site's real pages and their own metadata. And keep llms.txt in perspective — it's a cheap hedge, not a strategy. Whether AI assistants actually mention your brand when customers ask is a separate question, and it's the one our free AI Visibility Checker answers with live model probes.

Frequently asked questions

What does a valid llms.txt file look like?: Markdown with a fixed shape: exactly one H1 title as the first line, an optional one-line blockquote summary right after it, optional free-form context, then H2 sections whose contents are link lists in the form '- [Link text](https://url): short description'. An 'Optional' section at the end marks links agents may skip when they need a shorter context.
Do AI engines actually read llms.txt?: Mostly no, and we won't pretend otherwise. Ahrefs measured that 97% of published llms.txt files get zero requests from AI crawlers, and Google has said it doesn't use the file — John Mueller compared it to the meta keywords tag. Where it does get used is on-demand: coding assistants, agents, and dev tools fetch it when a user points them at your site.
If almost nobody reads it, why validate it?: Because the consumers that do exist fetch it programmatically and parse it strictly. A file with broken links, relative URLs, or an HTML error page served at the path is worse than no file — the agent wastes its context window on garbage or dead ends. If you publish one, it should be small, valid, and truthful.
What's the most common mistake this validator finds?: HTML served at the /llms.txt path. Many sites return their SPA shell or a soft-404 page with HTTP 200 for any unknown path, so tools that fetch llms.txt get a web page instead of markdown. The second most common: sitemap dumps — hundreds of bare URLs with no descriptions, which defeats the point of a curated index.
What's the difference between llms.txt and llms-full.txt?: llms.txt is a curated index: title, summary, and annotated links. llms-full.txt is the companion convention (popularized by Mintlify and Anthropic) that inlines the full text of your pages into one file. This validator checks the index file, and in URL mode also tells you whether an llms-full.txt exists at the same origin.

Related free tools

Generator

llms.txt Generator

Generate a well-formed llms.txt from your site's real structure and metadata — no AI-invented descriptions. Crawl, review, download.

Generator

llms-full.txt Generator

Turn your site into a single llms-full.txt — every page's readable content as one markdown file, ready to feed to any AI assistant.

Checker

AI Crawler Access Checker

Check which AI bots your robots.txt allows or blocks — ChatGPT, Claude, Perplexity, Gemini and more, with the exact matching rule.

Now find out what AI models actually say about your brand

These tools make your site readable to AI. The free AI Visibility Checker probes grounded AI models with live web search and shows the real answers, your score, and who wins instead of you — no login required.

Run the free AI Visibility Checker