Pillar 6 · Practice

Practice the exam, not just the content.

A 10-question diagnostic to find your weakest domain in 5 minutes, a 30-question mock with a full mental-model audit on the result page, and a 4-week study plan calibrated to your specific misconceptions. 402 vault-cited questions, 4 options each, priority-ranked distractors with mental-model labels.

Diagnostic free30-Q mock3 lifetime attempts720 pass mark

Start the free diagnostic Take the 30-question mock

Domain weights

Stratified by exam blueprint.

The mock samples each domain at the official CCA-F weight. Diagnostic uses an even split (2 per domain) since 10 questions can't represent a 27% / 18% / 20% / 20% / 15% breakdown without rounding artefacts.

Agentic Architectures

27%

Tool Design + Integration

18%

Agent Operations

20%

Prompt Engineering

20%

Context + Reliability

15%

10 minutes

Diagnostic

10 questions, 2 per domain. No login. Returns your weakest domain plus three concept links to read first. Best entry point if you're unsure where to start.

Start diagnostic

~45 minutes

30-Q mock

30 questions stratified D1 8 / D2 6 / D3 6 / D4 6 / D5 4. Scored to 1000 with 720 pass mark. Submit drops you on a result page with the full mental-model audit, concept hierarchy, and a 4-week study plan. 3 lifetime attempts.

Take the mock

FAQ

Frequently asked.

Is the diagnostic free?

Yes. The 10-question diagnostic is open to anyone. Two questions per domain, returns your weakest domain plus links to the concepts you should re-read first.

How is the 30-question mock different from the diagnostic?

The mock is the full evaluation: 30 questions stratified across domains by exam blueprint weight, scored to 1000, gated by a 720 pass mark, and produces a full mental-model audit + 4-week study plan after submit.

How many mock attempts do I get?

Three lifetime attempts on the 30-question mock, server-enforced. The diagnostic is unlimited.

What happens after a mock attempt?

You see a result page that audits the mental model behind your wrong answers. Each missed question maps to a concept hierarchy weighted by domain importance and how-wrong the distractor was. You get a 4-week plan with concrete next-actions.

Why 720 to pass?

Anthropic's CCA-F sets the pass mark at 720 of 1000. Same threshold here, applied to a 30-question scaled score. 22 of 30 correct lands at 733 and passes; 21 of 30 lands at 700 and does not.

What's the question bank size?

402 vault-cited V2 questions across 25 concept primitives and 14 architecture scenarios. Every question carries 4 options with priority-ranked distractors, the mental model under test, and per-option rationale.