Skip to content
CCAF Preparation

Domain 1 · 27% of exam · 16 of 60 items

Agentic Architecture & Orchestration practice questions

70 questions drawn from this domain’s 7 task statements, untimed and in order. Every option carries its own explanation, so a question you get right for the wrong reason still tells you something.

Domain 1 drill

Question 1 of 70

Scenario · Customer Support Resolution Agent

Your support agent returns an incomplete reply on roughly 8% of multi-step requests: it says "Let me look up your order" and stops there. The loop decides the agent has finished whenever the first content block of a response is text. What change would most effectively address this?

Getting value out of these

Read every explanation, including on the questions you answered correctly. The exam is a judgement test: the distractors are all things a working engineer has actually tried, and the reason each one fails is the material. A right answer picked for the wrong reason will not survive a differently worded question.

When you get one wrong, the useful question is not what the right answer was but whether you reached for a workaround over a mechanism — a bigger context window instead of less context, an instruction in the prompt instead of a constraint in the orchestration layer. That pattern repeats across the whole exam, and noticing it is worth more than memorising any single item.

These are untimed on purpose. Pace is a separate skill and belongs in the timed mock exam, where 60 items in 120 minutes gives you roughly two minutes each on questions that take real reading.

Where these questions come from

Each task statement’s lesson explains the material the questions test.

  1. 1.1Agentic Loops10 questions
  2. 1.2Multi-Agent Orchestration10 questions
  3. 1.3Subagent Invocation and Context Passing10 questions
  4. 1.4Workflow Enforcement and Handoff10 questions
  5. 1.5Agent SDK Hooks10 questions
  6. 1.6Task Decomposition Strategies10 questions
  7. 1.7Session State and Resumption10 questions