Paste any domain and this llms.txt validator fetches its /llms.txt, lints it against the spec, samples the links for 404s and hands you a letter grade with concrete fixes.
Just the domain is enough — the llms txt checker requests /llms.txt directly, exactly like an AI crawler would. That is also how to see the llms.txt file of a website yourself.
The llms.txt validator checks the H1 title, blockquote summary, section structure, link format, content-type and file size against llms.txt best practices.
You get an A–F grade, every issue ranked by severity, and a sample of linked pages tested for dead URLs.
The #1 llms.txt bug: your server returns an HTML fallback page instead of text. This llms txt checker spots it instantly.
Title, summary, sections, absolute links, descriptions — every rule from the llms.txt spec, checked in order.
A file full of 404s tells AI your site is stale. We fetch a sample of your links and flag every broken one.
Not just “valid/invalid” — a score, a letter grade and a prioritized fix list you can hand to a developer.
We also probe for llms-full.txt, the long-form companion file, and tell you whether it resolves.
Raw file preview plus a parsed outline — how to validate llm txt file structure without reading Markdown by squint.
How to see the llms.txt file of a website? Append /llms.txt to its domain — for example kairosy.ai/llms.txt — and the file renders as plain text. If you get your homepage or a 404 instead, the site fails the very first check of this llms.txt validator. It is a surprisingly common failure: single-page apps love to swallow unknown paths and return HTML.
The core llms.txt best practices are short: keep the file curated (dozens of links, not thousands), use absolute URLs, give every link a one-line description, serve it as text/plain, and keep it fresh — a stale file with dead links is worse than none. If you are unsure how to validate llm txt file changes after every deploy, bookmark this llms txt checker and re-run it; it takes about five seconds.
“Our llms.txt was silently serving the SPA shell for months. This caught it in one run — nothing else did.”
“The letter grade makes it easy to get eng buy-in. “We are a D” lands very differently than “please fix metadata”.”
“Dead-link sampling is the killer feature. Wish it checked every link, but the sample already caught three rotted URLs.”
“I run every prospect’s domain through it before a pitch. Instant conversation starter.”
Existence (a real 200, not an HTML fallback), content-type, the “# Title” heading, the “> summary” blockquote, “## Section” structure, link format and descriptions, file size, a dead-link sample, and whether llms-full.txt exists.
Type the domain followed by /llms.txt into your browser. If it renders as plain Markdown-ish text, it exists; if you see your homepage or an error, it does not.
Re-run this validator after each deploy, or add a simple CI step that fetches /llms.txt and fails when the response is HTML or non-200. The grade here maps cleanly to a pass/fail gate.
Curate rather than dump, use absolute links with one-line descriptions, serve text/plain at the site root, keep it small, and remove dead links promptly. Follow those and you will grade A here.
llms.txt helps AI read you; it does not force AI to recommend you. Run a free Kairosy scan to see what ChatGPT, Gemini, Claude and Perplexity actually say about your brand and where the gaps are.
Kairosy asks ChatGPT, Gemini, Claude and Perplexity the questions your buyers ask — and shows whether they recommend you, ignore you, or send buyers to a rival.
Run a free AI brand scan