Curious how this handles applets that need multi-step confirmation
before firing - for anything irreversible (locking a door, sending a
message to the wrong channel), I'd want Claude to explicitly confirm
before triggering, not just execute on the first ask. Would love to
hear directly from the team: is there a built-in confirmation layer
for higher-stakes actions, or is that left entirely to how the user
phrases their request?
Thanks for the thoughtful review, and sorry for the slow reply. Today an Applet runs its actions as soon as the trigger fires, with no confirm step. A confirm step for higher-stakes actions, like unlocking a door or sending a message, is one of our longest-standing feature requests. It's on our roadmap, and we want to get to it soon.