CCAR-P Advanced Agentic Architecture Practice Question
A Claude agent orchestrates three specialized subagents: one for data extraction, one for validation, and one for report generation. In production, the validation subagent sometimes receives malformed input from the extraction subagent and silently produces a passing result. You need the orchestrator to detect and contain these failures without halting the entire pipeline. Which design change is most appropriate?
⚠ Common exam trap
The trap here is trusting a failing subagent to accurately report its own health through confidence scores, instead of validating the data contract at the boundary.
Answer choices
Why each option matters
Answer the question above first, then reveal the full breakdown to understand why each option is right or wrong.
Correct answer & explanation
✓
Have each subagent return a structured result envelope containing a status, a schema-validated payload, and an error field, and make the orchestrator validate the envelope before routing onward.
Reliable multi-agent pipelines depend on explicit contracts at every handoff. A structured envelope with status, validated payload, and error field lets the orchestrator verify upstream output before trusting it downstream. That check enables targeted containment, such as retrying or quarantining one branch, rather than trusting a failing agent's self-assessment or retrying blindly.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✗
Increase the orchestration layer's retry count so that any subagent failure is retried several times before the pipeline reports an error to the caller.
Why it's wrong here
Retries help with transient faults but do nothing when the extraction subagent deterministically emits malformed data; the same bad input will fail the same way. Blind retries also multiply cost and latency. Without a validation contract at the handoff, the orchestrator cannot distinguish a transient error from a structural one, so retries alone cannot contain the failure.
- ✗
Merge the extraction and validation subagents into a single agent so that malformed intermediate data never crosses a component boundary.
Why it's wrong here
Collapsing the two roles removes the independent check that validation provides and hides errors inside one larger context. A single agent can still silently accept its own malformed output, and the failure becomes harder to isolate. The goal is a checked contract between stages, not the elimination of the boundary, so merging subagents reduces observability and containment.
- ✗
Instruct the validation subagent to always return a confidence score between zero and one, and have the orchestrator discard results below a fixed threshold.
Why it's wrong here
A self-reported confidence score is generated by the same subagent that is failing, so a malformed input can still yield a high score. Fixed thresholds are brittle across task types and provide no schema guarantees about the payload. The orchestrator would be trusting the failing component to report its own unreliability, which does not contain the fault.
- ✓
Have each subagent return a structured result envelope containing a status, a schema-validated payload, and an error field, and make the orchestrator validate the envelope before routing onward.
Why this is correct
A structured envelope with an explicit status and validated payload gives the orchestrator a deterministic contract to check, so malformed extraction output is caught before validation runs. Because the status is machine-readable, the orchestrator can retry, reroute, or quarantine just the failing branch instead of stopping the pipeline. This contains the fault at the boundary where it occurs.
About these practice questions
Courseiva writes every CCAR-P question from scratch — 262 in total, each with an explanation and a wrong-answer breakdown. None are copied from real exams or dumps. Learn why practice questions differ from exam dumps →
JA
Written and reviewed by Johnson Ajibi, MSc IT Security
Senior Network & Security Engineer · founder of Courseiva
Last reviewed September 2026 · checked against the official Anthropic exam blueprint
This CCAR-P practice question is part of Courseiva's free Anthropic certification practice question bank. Courseiva provides original exam-style practice questions with explanations, topic-based practice, mock exams, readiness tracking, and study analytics to help learners prepare for the CCAR-P exam.