At present, notifications can be set up for when an individual address is pwned or when an address on a domain someone is monitoring is pwned. This idea is to programatically enable calling an HTTP endpoint when this event occurs.

The benefit of this idea is that developers can then implement their own logic which is automatically invoked when an event like this occurs. It's of most value to those monitoring domains where notifications are no a rare occurrence, particularly when we're talking about larger domains.