CCAR-F Context and Reliability Practice Question
An architect is designing a Claude-based contract review tool. Because contract clauses are long and interdependent, the team plans to use extended thinking to improve reasoning quality. Which TWO practices should the architect follow to use this capability correctly? (Choose two.)
⚠ Common exam trap
The trap here is treating thinking blocks as reusable context or as a user-facing deliverable, when they are internal scratch work that should be sized appropriately and not fed back into later turns.
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
✓
Allocate a thinking budget that scales with task complexity and leave sufficient max_tokens for the final answer.
Extended thinking requires sizing the thinking budget to task complexity while preserving room for the final answer, and it requires not recycling prior thinking blocks into later turns. Together these keep reasoning thorough, cost proportionate, and multi-turn dialogues clean. Exposing thinking traces, maxing budgets unconditionally, and returning only thinking output all misuse the capability.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✗
Disable the final answer and return only the thinking output as the contract review result.
Why it's wrong here
The deliverable is the reviewed assessment, not the internal reasoning. Returning only thinking output omits the structured findings reviewers need and may expose unpolished, non-final content. The final answer is what should be validated against requirements and presented to users. Suppressing it defeats the purpose of the review workflow entirely.
- ✗
Set the thinking budget to the maximum on every request to guarantee the best possible clause analysis.
Why it's wrong here
Maxing the budget indiscriminately inflates cost and latency, and for simple clauses it yields no quality gain. Thinking budget should track complexity so routine checks stay fast and cheap. Always using the maximum also risks crowding the context and exceeding useful reasoning depth without benefit. Matching budget to task difficulty is the recommended practice, not a blanket maximum.
- ✓
Allocate a thinking budget that scales with task complexity and leave sufficient max_tokens for the final answer.
Why this is correct
Extended thinking consumes tokens for internal reasoning, so the budget must match the difficulty of the clause analysis and the max_tokens ceiling must still accommodate the visible response. Under-budgeting yields shallow reasoning, while a ceiling too low for the answer truncates output. Sizing both together keeps the reasoning thorough without starving the final contract assessment the user actually sees.
- ✗
Expose the raw thinking blocks to end users so they can audit the model's reasoning trace.
Why it's wrong here
Thinking content is not intended as a user-facing artifact and may be summarized or omitted depending on configuration. Presenting it as an auditable record can mislead reviewers and create compliance risk if it is incomplete. The audit trail should come from the final answer and citations, not from internal reasoning blocks. Exposing them also expands the surface for prompt-injection leakage.
- ✓
Strip or ignore prior thinking blocks when continuing a multi-turn conversation.
Why this is correct
In multi-turn use, earlier thinking blocks should not be fed back as ordinary context; they are internal and reusing them can distort subsequent reasoning and waste tokens. The model should reason afresh on the current turn using the conversation and any preserved answer content. Discarding stale thinking blocks keeps the dialogue coherent and avoids compounding noise across turns.
About these practice questions
One of 271 original CCAR-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 CCAR-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 CCAR-F exam.