Crawler access
Whether your robots.txt lets GPTBot, ClaudeBot, PerplexityBot, Google-Extended and the rest through. Points are split across every bot, so a blanket block costs the lot. This is the gate — fail it and nothing else counts.
Can AI find you, read you and quote you?
Search increasingly happens inside an answer, not a results page. This scan scores whether AI crawlers can reach your content, parse it, and cite you as the source.
Free
No signup
Results in 60 seconds
The six things we score
One hundred points, weighted by how much each one actually decides whether you appear in an answer.
Structured data and metadata
Valid JSON-LD, a real meta description, Open Graph tags and a meaningful title. This is what lets a model state what your page is about rather than infer it.
Rendered content
How much readable text exists in the raw HTML before JavaScript runs. Many AI crawlers do not execute JavaScript, so a client-rendered page can look empty to them while looking fine to you.
Semantic HTML
A single h1, real landmark elements, and enough heading structure to give the page an outline. Structure is how a model decides which passage answers which question.
Discoverability
An XML sitemap, an llms.txt file describing your site to language models, and valid EntityMap markup. The cheapest points on this list and the most often missing.
Performance
Whether pages respond quickly enough to be fetched reliably at crawl time. Crawlers give up faster than people do.
Why this is suddenly urgent
The referral model that SEO was built on is being replaced by answers that cite a handful of sources. Being one of them is the new ranking.
Answers are replacing clicks
AI Overviews, ChatGPT search and Perplexity answer the question in place. If your content is not readable by those systems you are absent from the answer, not merely ranked below it.
Most sites block AI crawlers by accident
A robots.txt rule copied from a template, or a CDN bot filter switched on for scrapers, is usually all it takes. Nobody notices, because nothing breaks visibly.
Client-side rendering makes you invisible
If your content only exists after JavaScript runs, crawlers that do not execute it see an empty shell. This scan tells you what they see, not what you see.
How the scan works
We look at your site the way a language model crawler does.
-
We check the gate first
Your robots.txt is parsed per AI user-agent, so you find out exactly which bots are allowed and which are turned away.
-
Pages are fetched raw and rendered
We compare the raw HTML against the rendered page, which is what exposes content that only exists after JavaScript runs.
-
You get a score and a fix list
Six category scores adding to 100, each with the specific gap behind it — a missing sitemap, invalid JSON-LD, a blocked crawler.
AI-readiness questions
What the score means, whether to allow AI crawlers at all, and what llms.txt is for.
How is the AI-readiness score calculated?
One hundred points across six categories: crawler access 30, structured data and metadata 20, rendered content 15, semantic HTML 15, discoverability 10 and performance 10. Crawler access carries the most weight because it is a gate — content a model cannot fetch cannot be cited no matter how well it is marked up.
Should I actually allow AI crawlers to use my content?
That is a business decision and the scan does not make it for you. If you want to appear in AI answers, you have to be readable by the systems producing them. If your content is the product — paid research, licensed material — blocking is legitimate, and you should expect a low score here and not care. Note the crawlers are not interchangeable: some fetch for training, some only when a user asks a question in real time, and robots.txt lets you distinguish them.
What is llms.txt and do I need one?
It is a plain-text file at the root of your site, in the same spirit as robots.txt, that tells language models what your site is and points at the pages worth reading. It is a young convention rather than a ratified standard, but it costs an hour to write and it is a direct, unambiguous statement of what you want quoted. OmniCheck publishes one itself.
Is this the same thing as SEO?
It overlaps but it is not the same. Classic SEO optimises for ranking in a list of links. This optimises for being parsed and cited inside a generated answer, which puts far more weight on machine-readable structure — valid JSON-LD, clean heading outlines, content present without JavaScript — and no weight at all on things like backlink profiles.
My site scores badly on rendered content. What does that mean?
It means there is little readable text in your HTML before JavaScript executes. Server-side rendering, static generation, or prerendering for crawlers all fix it. The quickest way to see the problem yourself is to view source, rather than inspect the live DOM, and look for your actual copy.
Which AI crawlers do you check for?
The scan checks the current named agents from OpenAI, Anthropic, Perplexity, Google, Apple, Meta, ByteDance, Common Crawl, Cohere and Amazon, plus Bingbot, which feeds AI answers. The list is configuration rather than code because bot names change faster than releases do.
Check the rest of your site health
AI-readiness is one of five checks. Each runs on its own, or all of them together in one score.
Accessibility
Find the barriers that turn visitors away, and the compliance gaps that carry legal risk.
Speed
Slow pages lose sales and rankings. See exactly which pages drag, on mobile and desktop.
Broken Links
Dead ends frustrate visitors and quietly drain SEO. We crawl every page and flag every 404.
Security
Passive checks on transport security, defensive headers and public exposure signals.
Monitoring
Know your site is down before your users tell you, and watch its health score over time.
Full Site Check
Run every check in one pass and get a single health score with the top three things to fix.