It only reads
Every request is an HTTP GET. It cannot submit a form, sign in, add anything to a cart, or buy anything, because it holds no tool that can. A task that says "reach the checkout" ends at the checkout page's URL.
If you found this in your access logs, it was us. AXRushAgent is an agent we send to a website on behalf of the person who owns it, to find out whether an AI agent could complete a real task there. It arrives only on sites somebody added to AX Rush and asked us to watch.
Every request is an HTTP GET. It cannot submit a form, sign in, add anything to a cart, or buy anything, because it holds no tool that can. A task that says "reach the checkout" ends at the checkout page's URL.
It reads your robots.txt before its first page and honours the rules for its own name, falling back to the wildcard group. A path you disallow is never requested, and the owner's report says the task was blocked and why.
A run is capped at a dozen pages and under a minute, one request at a time per host, with a pause between them. Several samples of the same task share one cache, so five measurements usually cost your server one or two requests.
Add this to your robots.txt. It takes effect on the next run, and no further page is requested.
User-agent: AXRushAgent Disallow: /
If you are the site's owner and you would rather it stopped entirely, removing the site from AX Rush ends it immediately.
Write to info@axrush.com with the timestamps from your logs and we will tell you exactly which run they belong to.