Learn by reasoning through why answers are right or wrong. Every question shows the winning logic, the traps, and the principle to carry into the next question.
60questions
120 mintime limit
720 / 1000to pass
4 of 6scenarios
5content domains
1 correctof 4 options
★ Focus areas. Everything tagged ★ — questions, principles, traps, playbooks, and reference tables — was added after the 8/12/2026 attempt (702/720) and targets only the objectives that scored below 80%. Use the ★ Focus areas chip on the Worked Questions tab to drill just those. Nothing from the original guide was removed.
New · 9/12. A smaller follow-up from reader feedback — fork vs resume, built-in tool selection, MCP loading, the retry boundary, and plan mode vs direct vs interview. Those items carry a New · 9/12 tag, and the matching chip drills just those 13 questions.
How to use this guide. The single answer is rarely the hard part — the incorrect options are written to look right to someone with partial knowledge. So for each question, read why the three distractors fail. They almost always fail for the same handful of reasons, collected in the Principles and Traps tabs. Master those and you can reason to the answer even on a question you've never seen.
Worked Questions
Filter by domain, click a question to expand. Green = correct with why; red = the trap and why it fails.
All domains
D1 · Agentic Architecture
D2 · Tool & MCP
D3 · Claude Code
D4 · Prompt & Output
D5 · Context & Reliability
★ Focus areas (weak objectives)
New · 9/12
Expand all ▾
Decision Principles
These recurring rules decide most questions. When stuck, run the answer options past the one that fits the situation.
Traps & Exceptions
Wrong answers that look right, plus the one condition that flips a normally-correct approach into the wrong one.
Task Recall
High-level build playbooks — the steps to actually do each thing the exam covers, from setup through verification. Knowing the sequence lets you spot the missing or out-of-order step in an answer.
Visual Map
The shapes behind the questions — what a request actually contains, how the loop turns, where hooks bite, and how the decision trees branch. Skim these before a practice run to load the structure, then let the questions test it.
Quick Reference
The exact flags, paths, and configuration values that show up as distractors when they're slightly wrong.