Robots.txt checker for AI crawlers
See which AI crawlers can read your site and which cannot. It separates the ones that come to train from the ones that come to answer, because they are not the same thing.
Training and answering are not the same thing
GPTBot collects pages to train models. OAI-SearchBot is the one that goes looking when ChatGPT answers with search turned on. Two different bots, from the same company, with two purposes that have nothing to do with each other.
Half the internet copied a robots.txt that blocks both at once, believing it only closed off training. The result is a site that still does not fully control whether it gets used for training, because not every crawler respects the file, and that on top of it can no longer be cited. The worst of both options.
What this check does and does not do
It downloads your domain's robots.txt and works out, rule by rule, whether each crawler may request the root of your site. It applies the real rules: the most specific group wins, groups do not stack, the longest rule wins, and a tie goes to Allow.
What it does not do: check whether the crawler respects the file. A robots.txt is a request, not a wall. The crawlers from OpenAI, Anthropic, Google and Perplexity respect it and document that they do; others do not, and for those you need to block by user agent at the server.
Being allowed is not being cited
A crawler being able to get in is necessary but not sufficient. Opening up robots.txt does not make a model name you: it makes it possible. What decides whether it names you is whether your page answers the question better than the others, and whether your entity is defined well enough for the model to know who it is talking about.
That is why this is a way of ruling things out, not a diagnosis. If everything comes back blocked, you know why you do not show up. If everything comes back allowed, the reason is somewhere else and has to be measured.
This tells you whether they can read you. Not whether they name you.
The three checks here rule things out: they tell you whether something is stopping you from appearing in an answer. Whether you actually appear, and who appears instead of you, has to be measured by asking the models. The scanner does that free in about a minute.