Debugging a failed multi-agent run is hard because prompts, token counts, and tool calls are scattered without a single timeline. What should you implement in Microsoft Foundry?
Select an answer to reveal the explanation.
Short Explanation
C is the Foundry tracing skill: tokens, prompts, correlation IDs, alerting, and execution tracking. Cross-agent debugging needs a correlated timeline of reasoning and tools. A disabling tracing saves nothing when incidents drag on. B status codes alone lack agent forensics. D browser localStorage is not enterprise observability. Correlate orchestrator and specialist spans with one correlation ID per workflow.
Full Explanation
Correct Answer — C
Tracing in Foundry should include tokens, prompts (subject to privacy policy), correlation IDs, alerting, and execution tracking so multi-agent runs can be reconstructed and debugged end to end.
Why A is wrong: No traces make production debugging impractical.
Why B is wrong: HTTP codes omit prompt/tool reasoning detail.
Why D is wrong: Client localStorage is not a durable observability backend.
Exam tip: Correlation IDs stitch multi-agent hops into one trace.