The production rung: CCAR-P

Foundations asked can you design a working Claude system? Professional asks can you own one in production — ship it, measure it, govern it, defend it to stakeholders, and keep it healthy through model upgrades. This track turns your Foundations capstone into that system, one production layer per week.

standalone items
(MC + multiple-response)

120 min

≈ 1.9 min / question

720/1000

scaled pass ≈ 46 of 63

$175

Pearson VUE, via Anthropic Partner Academy

Seven domains — and where a Foundations graduate leaks marks

Roughly two-thirds of this exam is Foundations material at production altitude — scale, failure modes, and money. The other 35% is new territory (marked below): governance, stakeholder work, and team enablement. That's where to spend your first study hours.

Domain weights

Purple = absent from Foundations. Integration (19%) is the single biggest domain — and the most buildable: week 2 of the plan is an access-controlled RAG pipeline.

The path

Read the blueprint live
All 7 domains decoded — what each tests, how it maps to Foundations, and the pro-level trap in each. Compiled from two independent summaries of the official exam guide that agree on every fact; verify against Partner Academy before booking.
Hold the Foundations floor live
Domains 1–4 lean on the Foundations track: 26 explainers, the 60-question timed mock, and the Bench. At Professional altitude the mock should feel easy — if it doesn't, the leak is downstairs.
Run the 6-week production-hardening ladder live
One production layer per week onto your capstone: failure story → access-controlled retrieval + tracing → regression suite + cost dashboard → governance pack → SLOs + exec brief → the paved road. Every layer is an evidence-pack artifact.
Clear the six Professional Bench stations live
Produce and grade the ladder's actual evidence: a failure-aware architecture dossier, authorization + trace contract, release scorecard, governance pack, lifecycle brief, and team paved-road manifest. Drafts and the latest grading evidence stay in your browser, with no API key.
Drill all seven domains, then sit the timed mock live
Seven "Professional ·" drill sets (74 questions) covering every CCAR-P domain, plus a timed 63-question mock at the real blueprint weights — 120-minute clock, feedback withheld to the end, per-domain breakdown. 46/63 ≈ the passing zone. Mixed formats like the real thing: single-answer items plus eight multiple-response items, scored all-or-nothing.
Drive the seven interactive pro explainers live
Seven measured simulations in the house style: access-controlled RAG (where privacy is zero by construction), A/B testing mechanics (assignment, uncertainty, value and guardrails), cost optimization (quality floors and cost per successful task), the enforcement map (one policy, three homes, measured), GDPR, HIPAA & FedRAMP (different scope, different architecture), SLOs & error budgets (same month, different promises), and the model-upgrade pipeline (pin → shadow → stage → rollback, counted in users hurt).

⚠️ Honesty note: exam facts here are compiled from third-party summaries of the official guide (they agree with each other on every domain, weight and format fact) — check the current official exam guide on Anthropic's Partner Academy before booking, since weights get revised between exam versions. All questions on this site are study aids written for this track, not real exam items.