CCAO-F Prompting and Context Engineering Practice Question
A support-engineering team is designing a Claude prompt to triage incoming bug reports into one of five severity levels. They observe that when the report is ambiguous, Claude sometimes invents a justification for a severity that is not actually supported by the text. They want the prompt to make uncertainty explicit rather than forcing a confident label. Which TWO changes should they make to the prompt? (Choose two.)
⚠ Common exam trap
The trap here is believing that more few-shot examples or step-by-step reasoning alone will fix fabrication, when the real fix is giving the model an allowed way to say it does not know and requiring it to quote its evidence.
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
✓
Add an explicit 'unknown' or 'needs_review' option to the allowed severity set and instruct Claude to choose it when the report lacks sufficient detail.
Grounding each severity in a verbatim evidence quote forces the model to anchor claims in the report text, and offering an explicit needs_review category gives it a legitimate way to express uncertainty instead of inventing support. Together these two changes convert an overconfident classifier into one that surfaces ambiguity for human follow-up.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✓
Add an explicit 'unknown' or 'needs_review' option to the allowed severity set and instruct Claude to choose it when the report lacks sufficient detail.
Why this is correct
Forcing a choice among five severities leaves no honest escape hatch, so the model confabulates. Adding a needs_review category gives Claude a legitimate output for under-specified reports, which is exactly the behavior the team wants. It converts a hallucination pressure into a routing decision that humans can resolve.
- ✓
Instruct Claude to output a separate 'evidence' field that must quote the exact phrase from the report supporting the chosen severity.
Why this is correct
Requiring a verbatim evidence quote ties each severity label to text actually present in the report. If no supporting phrase exists, the model cannot fabricate a quote without it being visibly absent, which surfaces ambiguity. This grounding technique reduces unsupported severity assignments and gives reviewers a checkable audit trail for each classification decision.
- ✗
Raise the temperature so Claude explores a wider range of possible severity interpretations for each report.
Why it's wrong here
Higher temperature increases sampling diversity, which amplifies the tendency to produce varied and less-grounded justifications. For a classification task where the team wants conservative, evidence-backed labels, raising temperature works against the goal and would likely increase fabricated support for severities that the report does not warrant.
- ✗
Ask Claude to think step by step silently and return only the final severity label with no supporting text.
Why it's wrong here
Suppressing all supporting text removes the observable link between report content and label, making it harder for reviewers to detect invented justifications and impossible to audit. It also does not give the model an outlet for uncertainty, so ambiguous reports still get a confident label with no trace of the reasoning.
- ✗
Provide ten few-shot examples in which every report is confidently assigned one of the five severities.
Why it's wrong here
Demonstrating only confident assignments teaches the model that a definite severity is always expected, reinforcing the exact confabulation the team is trying to remove. Few-shot examples should include ambiguous cases labeled needs_review so the model learns that abstaining is an acceptable and demonstrated behavior.
About these practice questions
One of 259 original CCAO-F practice questions on Courseiva, each with a full explanation and wrong-answer analysis — not exam dumps or protected exam content. 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 CCAO-F 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 CCAO-F exam.