Free AI & SEO tool
Before ChatGPT, Claude or Perplexity can cite your page, their bots have to reach it and read it. Enter a URL: the checker tests the page as each AI bot, shows what AI sees without JavaScript, and tells you which of the new agent standards matter, and which don't yet.
Most agent readiness scores add up everything, including standards nobody uses yet. This one scores only what decides whether AI can reach and read your page. Three things matter most:
If robots.txt or a firewall blocks the bot a platform uses to fetch pages for answers, that platform can't cite you. Since 1 July 2025, Cloudflare blocks AI crawlers by default on new domains, so sites can block them without knowing.
A study by Vercel and MERJ found that the major AI crawlers, from OpenAI, Anthropic and Perplexity, don't run JavaScript. Content that only appears after scripts run is invisible to them.
Blocking GPTBot or ClaudeBot stops model training, not AI search. Blocking OAI-SearchBot or Claude-SearchBot keeps you out of the answers. Many sites mix them up.
Whether the site allows the bots that fetch pages for AI answers (OAI-SearchBot, Claude-SearchBot, PerplexityBot and their user-request versions), the training bots, and Google and Bing.
The page is fetched as a normal browser and as each AI bot. If the browser gets the page and a bot gets an error or a security check, a firewall is likely turning that bot away.
Most AI crawlers don't run JavaScript. The checker reads the HTML as it arrives and shows the text AI would get.
Noindex, the canonical tag, redirects, response time, page size, title, main heading and language.
The checker pretends to be each bot. Real bots can be treated differently, because some firewalls also check where a request comes from, so a block found here is likely, not certain. Check your server logs to be sure.
These are new and don't count towards the score: none of them has been shown to get pages cited more often. What is known about each is below.
A line that states whether AI may use your content for search, answers and training. A voluntary preference: no AI company has confirmed it honours it.
A plain-text version of the page for agents that ask for it. Claude Code and some other AI tools request it; the big AI crawlers don't, and no effect on citations has been shown.
A summary file for language models. Google says you don't need it, and AI crawlers rarely request it.
A standard way to point agents to machine-readable resources in the response headers. No AI crawler has been shown to use them.
Other agent readiness scanners also fail sites on standards that only apply to some businesses. For most company and content websites, these don't matter:
For webshops that want AI agents to buy on their behalf. ACP and UCP also need a product feed and approval through OpenAI's or Google's merchant programmes.
For sites that offer an API or a service agents log into or call.
Lets a page offer actions to AI agents in the browser. So far only in Chrome, behind a trial.
DNS records that advertise agent services. Still a draft proposal, not an adopted standard.
For most websites: AI bots can reach the page, read its content without running JavaScript, and nothing tells them to stay away. That decides whether ChatGPT, Claude, Perplexity or Google's AI features can use and cite the page. The newer agent standards come after that.
It fetches your page with each bot's user agent. Some firewalls also check whether a request really comes from that company's servers. A real bot can then be let in where the test was refused, or the other way round. Your server logs show what real bots received.
That is your choice. They collect content to train models; blocking them doesn't stop the bots that fetch pages for AI answers. But models answer many general questions without searching, from what they learned in training. If you want them to know your brand, allow the training bots too.
Not to be cited. Google says you don't need new machine-readable files for its AI features, and none of the big AI crawlers has been shown to use llms.txt. Markdown is requested by some AI coding tools, such as Claude Code, so it can help agents, but no effect on citations has been shown.
Being reachable is only the first condition. In my research, how closely a page matched the question was one of the strongest factors in being cited, and most AI citations came from pages outside Google's top 10.