Hey Eugen — thanks for the suggestion and the screenshot. this is a good idea.
For context, that “Waiting” countdown is the agent’s own scheduled wait: when it starts something slow (a CI run, build, or long command), it estimates how long that should take and waits before checking back in, so it isn’t constantly polling. The length is chosen per-operation and isn’t capped on purpose, since some tasks genuinely need to run long.
But you’re right that if you can already see it’s finished, sitting through the rest of the timer is wasted time. We’ll leave this thread up for other folks to weigh in and share their thoughts!