CCAR-P Advanced Agentic Architecture Practice Question
When designing an agent that must perform highly sensitive operations (e.g., deleting records), what architectural pattern is mandatory?
⚠ Common exam trap
Candidates often suggest relying on model-level security or fine-tuning to prevent errors, failing to recognize that human-in-the-loop is the only mandatory safeguard for high-stakes, irreversible actions.
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
✓
Implement a mandatory human-in-the-loop confirmation step.
A 'Human-in-the-loop' (HITL) gate for sensitive actions is non-negotiable for enterprise safety. The agent generates the plan and the proposed tool call, but the system architecture must intercept this and require explicit human verification before execution. This prevents accidental data loss or malicious exploitation of the agent's capabilities, balancing autonomous productivity with critical risk management in production environments.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✗
Fine-tune the model to never delete records.
Why it's wrong here
Fine-tuning is not a reliable safety control; it can be bypassed through prompt injection. You need an explicit, architectural, and programmatic 'gate' that requires a human sign-off. Safety constraints must be enforced at the system/code level, not just by hoping the model's training keeps it in line.
- ✓
Implement a mandatory human-in-the-loop confirmation step.
Why this is correct
The HITL pattern forces an external, verifiable authorization layer into the workflow. This ensures that no destructive or sensitive action can proceed without an explicit, audit-trailed human approval, which is the gold standard for secure agentic systems dealing with sensitive backend operations and data integrity.
- ✗
Use a low-temperature setting for sensitive tool calls.
Why it's wrong here
Lowering temperature does not prevent an agent from deciding to perform a destructive action; it just makes it more consistent in that decision. If the agent's logic is flawed, a deterministic model will just execute the destructive act with high confidence, which actually makes the situation worse.
- ✗
Log the action to a secure bucket after execution.
Why it's wrong here
Logging is for post-mortem analysis, not for preventing the action in the first place. For sensitive operations, you need a preventative mechanism, not a reactive one. Logging the destruction of data after it's already gone does not mitigate the damage, making this an insufficient architectural choice.
About these practice questions
This CCAR-P question is part of Courseiva's 262-question bank — original exam-style content with full explanations and wrong-answer analysis, never real exam questions or exam 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.