CCA-F Question 17

Community Discussion

Written by other learners. Comments can disagree with the marked answer — check the official documentation when in doubt.

• Chose A1

  The citations already exist at the subagent level. The failure happens during synthesis, where provenance is lost. A structured contract keeps provenance attached to each claim throughout the pipeline,

• Chose A1

  Structured claim-source mappings treat citations as first-class data fields, not text annotations: { "claim": "Global market size is $35B", "confidence": "high", "sources": [{"doc": "peer-review-2024", "page": 12, "url": "..."}] } When the synthesis agent receives these from both agents, it's required to preserve and merge the source arrays — not rewrite prose hoping markers survive. The citation trail is machine-readable and explicitly part of the data contract.

• Chose A1

  Why A is best: Each subagent outputs findings as structured claim → source mappings. Provenance travels alongside every claim through the entire pipeline. The synthesis agent merges both the findings and their associated attribution metadata. Citations become deterministic rather than inferred after the fact. This follows the principle that attribution should be preserved throughout the workflow, not reconstructed later. Don't try to reconstruct provenance after synthesis. Preserve provenance as structured metadata throughout the pipeline. That is exactly what A achieves