AI search check
The AI search check tells you whether the crawlers behind ChatGPT search, Claude, Perplexity, Copilot and Google's AI answers can reach your page and read it.
Use it when your product never comes up in AI answers, or after you change robots.txt or your hosting.
Blocking AI training crawlers is your choice and never lowers the score; blocking the search crawlers does.
Built by the IndieHacked team
Enter an address and run the check. The score and what to fix show here.
How the check works
What it reads
The check fetches your page without running JavaScript, as the AI crawlers that skip it do, plus /robots.txt, your sitemap and /llms.txt. It scores five topics out of 100 and averages them.
AI crawler access
robots.txt must let the search and answer crawlers reach the page: OAI-SearchBot for ChatGPT search, Claude-SearchBot for Claude, PerplexityBot, Googlebot for AI Overviews and AI Mode, bingbot for Copilot and Applebot. Each company documents these separately from its training crawlers. Blocking GPTBot, ClaudeBot, Google-Extended or CCBot keeps your pages out of training and never lowers the score. If your host's bot protection shows our checker a challenge, the check says so and scores nothing: hosts such as Vercel let verified AI crawlers through, and our checker isn't verified yet.
Readable without JavaScript
Google renders JavaScript and still recommends server rendering, because not every crawler does. OpenAI, Anthropic and Perplexity publish nothing on it; Vercel's study of its own traffic in December 2024 found their crawlers fetching JavaScript without running it. Under 50 words in the served HTML is a problem, under 150 one to improve.
Answer first, who you are, snippet controls
A title, a meta description and an opening paragraph in the HTML, which Bing's guidelines ask for near the top. An Organization and a sitemap so the site ties to your name. No noindex, nosnippet or max-snippet:0, which Google applies to AI Overviews and AI Mode, and no NOARCHIVE or NOCACHE, which Bing applies to Copilot. llms.txt is reported and never scored: no major AI company says it reads the file.
Questions founders ask
I block AI training crawlers. Will that hurt my score?
No. The check scores only the crawlers that put pages into answers: OAI-SearchBot, Claude-SearchBot, PerplexityBot, Googlebot, bingbot and Applebot. Blocking GPTBot, ClaudeBot, Google-Extended or CCBot shows as information and leaves the score alone.
Does the check prove I'll be cited in ChatGPT or Claude?
No. It checks that the answer engines can reach and read the page, which they need before they can cite it. Whether they pick your page depends on what it says and what else answers the same question.
Do I need an llms.txt file?
It's optional. No major AI company says its crawlers read it, and Google says such files aren't needed for its AI features. The check notes whether you have one and doesn't score it.
What does the fix prompt do?
It turns the findings into instructions for your coding agent (Claude Code, Cursor, Codex, Lovable or Bolt), written for what your site is built with when the check can tell. Paste it in, review the changes, then run the check again. Fix prompts are part of Plus, free with Weekly Unpack: subscribe with your account's address and every prompt opens.
Why do I need an account to run a check?
Each check reads a live site and asks Claude to write the verdict, which costs money every time. An account lets us keep it free with a daily limit of 10 checks, and keeps your past runs so you can compare a new score with the last one.
What happens to the address and the results?
We store the address, the score, the findings and the verdict with your account so you can see them again. The page's text goes to Anthropic, which runs Claude, to write the verdict and the prompt; the copy check also sends the address to Firecrawl to read the page as a visitor sees it. Runs are deleted after 12 months, or with your account.
Report an error. Spotted a factual or language mistake? .