Can the assistants read your website?
Type a domain. We read it the way an AI assistant such as ChatGPT does: the home page as a browser, as ChatGPT-User, as GPTBot, as ClaudeBot and as PerplexityBot, plus robots.txt, llms.txt and the sitemap. A score out of 100 and the findings in plain words, in about ten seconds. No sign-up to see the score.
Reading your site as a browser, as ChatGPT-User, as GPTBot, as ClaudeBot and as PerplexityBot. Ten seconds or so.
Can each reader get the page?
Https answering 200, a fast first byte, and the same page for ChatGPT-User, GPTBot, ClaudeBot and PerplexityBot as for a browser. A block at the edge, a challenge page or a noindex tag shows here.
Is there something to read?
Words in the HTML without JavaScript, a title and description, a canonical address, an answer-first paragraph under the headline, a visible date and contact details. An empty app shell scores nothing here.
Does the site say what it is?
robots.txt letting the AI crawlers and the search crawlers in (reported as allow or disallow, never a verdict), Organisation and FAQ structured data, llms.txt, a sitemap with lastmod dates.
We fetch from our server with the reader's user agent set; rules that check a crawler's address may treat the real reader differently. Rate limits are per visitor address, hashed daily. Results are kept 90 days at a link with nothing personal in it.
About the check.
It is the first reading we take on every site we work on, and it is the same engine the plans run every month.
What exactly do you fetch?
Your home page five times, as a browser, as ChatGPT-User, as GPTBot, as ClaudeBot and as PerplexityBot, so an edge block shows; then robots.txt, llms.txt and sitemap.xml. Nothing else. Each fetch has an eight-second limit and a size cap.
Is it safe to run on my site?
Yes. It is five ordinary page requests and three small files, the same as a crawler would make. We fetch only public hostnames, we refuse private addresses, and we never store the page itself; the result keeps the status codes, the first 300 characters of visible text and the findings.
Why is the score not a verdict on my rankings?
Because it measures what an assistant can see and declare, not what it chooses. A high score means the assistants can read you and know what you are; being chosen also depends on mentions, reviews and how your page answers the question. Watch shows you the choosing.
What does "blocked at the edge" mean?
When we asked for your home page as a named AI crawler, your server or firewall answered differently than it did for a browser, usually a 403, a challenge page or a 429. A managed "block AI bots" switch often does this. If ChatGPT-User is blocked, ChatGPT cannot fetch your page to answer anyone.
Can I run it again?
Yes. Five checks an hour per visitor. The same domain is answered from the last result for 30 minutes, so change something, wait half an hour, and run it again.