Design an evaluation loop for an agent that learns from reviewer feedback without silently drifting.

Instruction: Describe how you would use reviewer feedback to improve an agent while keeping behavior stable.

Context: Assesses whether the candidate can design a practical architecture and explain the main tradeoffs. Describe how you would use reviewer feedback to improve an agent while keeping behavior stable.

Official answer available

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

I want reviewer feedback to improve the system, but not by leaking directly into the live prompt. The safer pattern...

Related Questions