CCAR-P Advanced Agentic Architecture Practice Question
A production agent uses a ReAct loop and frequently reaches its maximum step budget while still mid-task, then returns a partial answer that looks complete. Telemetry shows the agent often re-reads the same file and re-queries the same database row across consecutive steps. Which change most directly reduces wasted steps while preserving the agent's ability to finish?
⚠ Common exam trap
The trap here is treating the step budget as the problem, when the budget is only the boundary that exposes redundant tool calls as the real inefficiency.
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 a step-level deduplication cache that returns the prior result when the agent issues an identical tool call with identical arguments, and inject a reminder of remaining budget into the loop.
The wasted steps come from repeating identical tool calls, so the most direct fix is to detect and short-circuit duplicates by caching results keyed on the call and its arguments. Pairing that with visibility into remaining budget helps the model spend its steps on novel actions and wrap up cleanly. Adjusting the budget up or down changes when the agent stops but not how much of its work is redundant, and collapsing the loop removes the observation-driven adaptation the agent depends on.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✗
Raise the maximum step budget from 15 to 60 so the agent always has enough room to finish any task.
Why it's wrong here
Raising the ceiling masks the symptom but does not reduce the redundant reads and repeated queries that consume steps. The agent would burn more tokens and latency before hitting a higher wall, and a partial answer that looks complete is still possible if the budget is exhausted again. This defers the problem and increases cost without improving step efficiency.
- ✗
Replace the ReAct loop with a single-shot prompt that asks the model to plan every step up front and then answer without further tool calls.
Why it's wrong here
Removing the loop eliminates the ability to observe tool results and adapt, which is the core strength of a ReAct agent. For tasks requiring intermediate data, a single-shot plan cannot incorporate real outputs, so it either guesses or fails. This does not address redundant steps; it removes the iterative mechanism that makes tool use effective in the first place.
- ✓
Add a step-level deduplication cache that returns the prior result when the agent issues an identical tool call with identical arguments, and inject a reminder of remaining budget into the loop.
Why this is correct
Caching identical tool calls with identical arguments removes the repeated file reads and database queries that consume steps, directly attacking the observed waste. Surfacing remaining budget lets the model prioritize finishing over redundant exploration. Together they cut wasted steps while leaving the agent free to complete the task, rather than artificially capping its work or raising the ceiling without improving efficiency.
- ✗
Lower the maximum step budget to 8 so the agent is forced to be more decisive and avoid redundant work.
Why it's wrong here
Tightening the budget makes the agent hit the wall sooner, increasing the frequency of truncated partial answers. It applies pressure but provides no mechanism to stop duplicate reads or repeated queries, so the underlying waste remains and the agent now fails earlier. This trades one failure mode for a worse one without improving step efficiency.
About these practice questions
One of 262 original CCAR-P 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-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.