llms.txt validator
Free llms.txt Validator
Check an llms.txt file against the published spec. Paste the file or point us at a domain and get every error, warning, and note with the exact line number, plus the parsed outline showing how a model would read it.
Built for llms.txt validator
Validate llms.txt against the actual spec
Paste your file and this llms.txt checker parses it the way the spec defines it: exactly one H1, an optional summary blockquote, prose, then H2 sections holding markdown link lists. Every finding carries a line number, so you fix the file rather than hunt through it.
Paste mode runs entirely in your browser — nothing is uploaded, so an unpublished draft is safe. URL mode fetches https://yourdomain.com/llms.txt from our server and tells you the single most common failure outright: a site with no file that answers with its normal HTML page instead.
Common use cases
- Confirm a new llms.txt is spec-valid before you publish it
- Prove whether a live domain actually serves a file or a soft-404 HTML page
- Find relative URLs, duplicate links, and undescribed entries by line number
Validation runs entirely in your browser. Nothing is uploaded, so unpublished files are safe to paste.
What llms.txt is, briefly
llms.txt is a proposed markdown file published at the root of a site — yoursite.com/llms.txt — that gives a language model a short, curated map of your most important pages. It was proposed by Jeremy Howard of Answer.AI in September 2024. For the full picture — who reads it, what to put in it, and how to generate one — see the llms.txt hub. This page does one job: checking that a file you already have is correct.
What this validator checks
Findings come at three severities. Errors break the format: no H1, more than one H1, content sitting above the H1, a heading level other than H1 or H2 used structurally, a bullet inside a section that is not a - [name](url): description link, or a file that turns out to be HTML rather than markdown.
Warnings are things that parse but are almost certainly mistakes: relative URLs (the file is read with no base URL to resolve them against), duplicate links, empty sections, links with no description, a file or link count large enough to stop being a curated map, and — the one people miss — a section called ## Optional Resources rather than exactly ## Optional, which silently loses the “you may skip these” meaning the spec gives that one heading.
Info is context rather than a problem: links pointing off your own domain, the section and link counts, the file size, and whether an llms-full.txt is referenced.
The failure this catches that nothing else does
The most common real-world problem is not a syntax slip — it is a site that has no llms.txt at all and answers the request with its normal HTML page, often with a 200 status. Open it in a browser and you see something, so it looks fine. A model requesting it receives your page markup instead of a map. URL mode calls that out explicitly rather than trying to parse the HTML as markdown.
An honest note on what this buys you
Adoption of llms.txt is real among developer tools and documentation sites, but no major AI provider has publicly committed to reading it in their production answer surfaces, and Google’s own guidance is that you do not need special machine-readable files to appear in Search or its AI features. A valid llms.txt is cheap insurance and a useful exercise in deciding which pages matter — not a ranking lever. If you want the fuller argument, the llms.txt hub lays it out.
Keep going
- llms.txt hub — what the file is for, and a generator that builds one from your site.
- robots.txt guide — the file that actually decides whether AI crawlers may fetch your pages at all.
- Googlebot simulator — fetch any URL as GPTBot or ClaudeBot and see exactly what it receives.
- AI crawlers explained — which bots visit, what they take, and how to control them.
How to Use the llms.txt Validator
- 1Paste your llms.txt into the box, or switch to "Check a live domain" and enter your domain
- 2Read the pass/fail verdict at the top — errors break the spec, warnings work but are probably mistakes
- 3Work down the findings list; each one shows the line number and the offending line
- 4Check the parsed outline to see the sections and link counts a model would actually read
- 5Fix, re-paste, and re-run until the errors are gone and the warnings are ones you chose
Frequently Asked Questions
AI Search Guides
Connect this tool to the bigger AI-search workflow
Technical fixes and free tools help most when they support a stronger page program. These guides explain how AI-search visibility, answerability, and tracking fit together.
GEO
Generative Engine Optimization
The core GEO framework, case study, and operating model for AI search visibility.
AEO
Answer Engine Optimization
How to structure pages so answer engines can retrieve, trust, and cite them.
GEO vs SEO
GEO vs SEO
GEO vs SEO and SEO vs GEO: the difference between generative engine optimization and classic SEO, how much they overlap, and where AEO fits.
GEO SEO
GEO SEO
How generative engine optimization fits into modern SEO and why the overlap matters.
Related tools
If you searched for llms.txt validator, these tools solve adjacent formatting, technical SEO, and content-optimization tasks without leaving the workflow.
llms.txt Generator
Generate llms.txt with free strategy onboarding (sign-in required).
Robots.txt Generator
Create a robots.txt file for your website in seconds.
Googlebot Simulator
View any page the way Googlebot and AI crawlers see it.
Rich Results Test
Test any URL for rich snippet eligibility.
Schema Markup Generator
Generate JSON-LD schema markup for common page types.
Redirect Checker
Trace redirect chains and check HTTP status codes.
