An agent says "done". What do you know then? That he thinks he's done
That's why every ticket in my setup has a check that stands apart from the work. "The server is running" is a claim. "You can log in and NonStopWorker answers" is proof. And a check that is always green proves nothing: I break what it guards and see whether it turns red.
The more room an agent gets, the less you can lean on what he says himself.
The whole story is on my blog.