Stakeholders want a refund agent that can draft refund recommendations but must never execute refunds above $500 without human approval. How should you specify the agent’s design?
Select an answer to reveal the explanation.
Short Explanation
Correct answer: B. Architecture must state persona, scope, autonomy, and behavioral rules explicitly. Draft freely; execute high-value refunds only after human approval; tools enforce the threshold. Unrestricted DB admin (A) and full autonomy for all amounts (D) break control. Thresholds only in code comments (C) are not operable design. Make autonomy levels testable and visible in HAX flows.
Full Explanation
Design skills include specifying agent personas, scopes, boundaries, autonomy levels, and behavioral guidelines, plus human-in-the-loop for high-risk actions. Option B captures those elements and ties them to tool permission boundaries. Options A and D over-privilege the agent. Option C fails as a control specification. Document the autonomy matrix for stakeholders and implement approval workflows that emit audit events.