Prove you can put Claude to work reliably: evaluate outputs, design safe workflows, and pick the right product for the job.
Listed in domain order, with the published weight and task statements for each.
14% of the exam — Structuring instructions, examples, and constraints that hold up.
21% of the exam — Judging correctness, grounding claims, and catching silent failure.
12% of the exam — Matching surface and model to latency, cost, and accuracy needs.
16% of the exam — Fitting Claude into real processes with clear handoffs and review points.
12% of the exam — Projects, context, files, and keeping shared knowledge current.
15% of the exam — Policy, data handling, disclosure, and the limits of acceptable use.
10% of the exam — Diagnosing weak results and improving them methodically.
Work down the blueprint below and mark each published task statement as confident, shaky, or unknown. The exam writes its items directly against these objectives, so an honest self-assessment is the whole study plan in miniature.
Items are multiple-choice and multiple-response, and each one states how many responses to select. Every question is a realistic situation asking for the soundest judgement — practise choosing between four defensible-looking options, not reciting definitions.
Structure prompts, decompose tasks, and iterate to improve outputs across the task types the guide names: analysis, research, drafting, and brainstorming.
Configure a Project with instructions and knowledge sources, produce outputs, then evaluate them for accuracy, completeness, and bias. Configuration and Output Evaluation together carry a third of the exam.
Decide, quickly and repeatedly: is this data sensitive, is this an appropriate use case, and is this the point where a human must review or the work must be escalated to an Architect or Developer?
For each task, justify the Claude product feature and the model tier you would pick on cost, speed, and quality, and say how you would handle context limits — restart, summarise, or persist.
Try 40 free Associate questions with full explanations, then move on to the timed sets.