The Claude Certified Architect – Professional (CCAR-P) exam is the advanced tier above Foundations (CCA-F / CCAR-F). Where Foundations validates that you can make the right architectural choice for a scenario, Professional pushes into enterprise-scale work: integration architecture, evaluation and optimization, governance and risk, and the stakeholder and lifecycle discipline of running Claude systems in production.
This guide frames the exam and links out to a deep-dive for each of the seven domains. It's an independent, unofficial resource — not affiliated with, endorsed by, or authorized by Anthropic.
One important note on freshness: CCAR-P is brand new. Treat the mechanics below as they appear in the official exam guide, and always confirm specifics (fee, question count, eligibility) against Anthropic's official exam page before you book.
What CCAR-P is — and who it's for
CCAR-P targets experienced solution architects who design, integrate, evaluate, and govern Claude systems at organizational scale. There is no mandatory prerequisite — you can sit Professional without passing Foundations first — but the recommended background is several years of systems-architecture experience plus hands-on Claude/LLM production work. It assumes the Foundations material (agentic loop, MCP, structured output, context management) as a baseline and builds enterprise concerns on top.
Exam mechanics
| Parameter | Value |
|---|---|
| Exam code | CCAR-P |
| Format | 63 items · multiple-choice + multiple-response · 120 minutes |
| Passing score | 720 (scaled 100–1000) |
| Fee | $175 USD (partner-tier discounts may apply) |
| Delivery | Proctored — OnVUE online or Pearson VUE test center |
| Validity | 12 months → free renewal assessment, or full retake if lapsed |
| Eligibility | Claude Partner Network members |
The seven domains at a glance
The blueprint spans seven weighted domains. Integration is the heaviest; study time should follow the weighting.
| Domain | Weight | Focus |
|---|---|---|
| P1 — Solution Design & Architecture | 17% | Choosing the right design; agentic vs. workflow; tradeoffs at scale |
| P2 — Claude Models, Prompting & Context Engineering | 13% | Model selection/routing, caching, context engineering |
| P3 — Integration (RAG, MCP, Auth, Observability) | 19% | Retrieval, MCP servers, auth/secrets, tracing & metrics |
| P4 — Evaluation, Testing & Optimization | 16% | Evals, LLM-as-judge, cost/latency optimization |
| P5 — Governance, Safety & Risk Management | 14% | Guardrails, injection defense, audit, compliance |
| P6 — Stakeholder Communication & Lifecycle Management | 14% | Requirements, tradeoff comms, rollout, versioning |
| P7 — Developer Productivity & Operational Enablement | 7% | Claude Code at team scale, CI/CD, enablement |
How much of Foundations carries over
- Reused from Foundations: the agentic loop and
stop_reason, MCP tool/resource design, structured output andtool_choice, context management, and the deterministic-enforcement-vs-prompting mental model all reappear — mostly inside P2 and P3. - Net-new at Professional: evaluation frameworks and LLM-as-judge (P4), governance/compliance/injection-defense at scale (P5), and the stakeholder-communication and lifecycle discipline (P6) — these have little Foundations overlap and deserve the most fresh study.
- Deeper, not just wider: P1 revisits agent design but demands enterprise-grade reasoning about cost, latency, reliability, and reversibility, not just "which mechanism."
A study approach
Key takeaways
- 01CCAR-P is the advanced tier above Foundations — enterprise integration, evaluation, governance, and lifecycle.
- 0263 items, 120 minutes, 720/1000 to pass, $175, seven weighted domains; Partner Network eligibility.
- 03No mandatory prerequisite, but Foundations knowledge is assumed as a baseline.
- 04Integration (19%), Solution Design (17%), and Evaluation (16%) are over half the exam.
- 05The net-new study areas vs. Foundations are Evaluation (P4), Governance (P5), and Stakeholder/Lifecycle (P6).
- 06CCAR-P is brand new — confirm fee, format, and eligibility on the official page.
Common mistakes
Assuming Professional is just "harder Foundations questions."
It adds whole domains (evaluation, governance, stakeholder/lifecycle) that Foundations barely touches. Study those fresh.
Over-indexing on prompting and under-preparing integration and evaluation.
P3 (19%) and P4 (16%) are the two biggest technical domains. Build and instrument a real RAG+MCP system.
Trusting an unofficial site's fee/question numbers.
Use the official exam guide (63 items / $175 / 7 domains) and confirm on Anthropic's page.
Frequently asked
Do I need to pass Foundations (CCA-F) before Professional?
No — there's no mandatory prerequisite; you can sit Professional directly. But Foundations knowledge is assumed as a baseline, and Foundations-before-Professional is the recommended progression.
How is CCAR-P different from CCA-F?
Foundations validates picking the architecturally-correct mechanism for a scenario. Professional adds enterprise concerns — integration architecture (RAG/MCP/auth/observability), evaluation and optimization, governance and risk, and stakeholder/lifecycle management — and expects deeper tradeoff reasoning at scale.
Is CCAR-P widely available?
It's brand new and delivered through the Claude Partner Network (Pearson VUE). Confirm eligibility and any team-size requirements on Anthropic's official page before planning around it.
Independent, unofficial study material. Not affiliated with, endorsed by, or authorized by Anthropic. Every example is original and written to teach the public exam objectives — no real exam questions are reproduced. Technical details reflect Claude, the Anthropic API, Claude Code, and MCP as of July 25, 2026; always confirm specifics against current official documentation.
Test yourself
Turn what you just read into answers you can check.