A shell command the agent wants to run could be useful or destructive. How would you decide?

Instruction: Explain how you would judge whether to allow a risky command.

Context: Tests how the candidate diagnoses the problem, chooses the safest next step, and reasons through recovery. Explain how you would judge whether to allow a risky command.

Official answer available

Preview the opening of the answer, then unlock the full walkthrough.

I would evaluate it against scope, necessity, reversibility, and environment risk. If the command changes state broadly, touches external systems, or could destroy data, it should move into a stricter policy tier or require human review.

I also want to know why the...

Upgrade to view official answer

Related Questions