Free practice · multiple-choice · medium
Practice
Review a tool-action plan
Which details must be visible before approving an agent pilot?
Human-control stack
External action requires stronger control than an internal suggestion
- GoalBound allowed and prohibited outcomes
- AccessGrant least tools and data
- TestExercise normal, boundary, and failure cases
- ApprovePlace checkpoints before consequential action
- RecoverLog, stop, contain, and reverse where possible
Least permission, pre-action checkpoints, limits, logs, stop conditions, and recovery keep a tool-using system inside its approved boundary.