CCAR-F asks whether you can architect a Claude-based system. CCAR-P asks whether you can own it — the whole lifecycle, from the first discovery conversation with a stakeholder through RAG pipeline design, evaluation at production scale, regulatory compliance, and the ADRs that let someone else defend your decisions after you've moved on. It's Anthropic's most advanced certification: a longer exam (63 items vs. 60), a higher fee ($175 vs. $125), and domains centered on integration, evaluation, and governance rather than agentic architecture alone.
We just shipped the free Study Track for it — 25 lessons plus a capstone, structured the same way as our CCAR-F track: problem → concept → decision framework → a hands-on lab → a 6-question checkpoint quiz, no separate quiz page to navigate to.
What's in it
CCAR-P's curriculum overlaps with CCAR-F on purpose — a lot of the underlying decisions (model selection, prompting, tool contracts, MCP design) are the same skill, just reasoned about at a higher level of ownership. So 16 of the 25 lessons are shared with CCAR-F, and 9 are exclusive to the professional track:
| Exclusive lesson | What it's about |
|---|---|
| Business discovery, requirements & SLAs | Structuring discovery before writing a single line of architecture |
| End-to-end architecture & value trade-offs | Architecture as a sequence of declared trade-offs, not a diagram |
| RAG, retrieval & data pipelines | RAG as a data-shape problem, not a model problem |
| Integration protocols, identity & least privilege | Choosing an integration protocol by trust boundary, not convenience |
| Production observability, latency & cost | Turning an incident into a diagnosis instead of a guess |
| Enterprise governance, compliance & HITL | Compliance as an architectural constraint, not a final checklist |
| Stakeholder communication, ADRs & lifecycle | A decision nobody wrote down can't be defended later |
| Architect Professional capstone | Defend a complete system, from discovery to operation |
Domain coverage across all 25 lessons follows CCAR-P's official exam weights:
| Domain | Weight |
|---|---|
| Integration | 19% |
| Solution Design & Architecture | 17% |
| Evaluation, Testing & Optimization | 16% |
| Governance, Safety & Risk Management | 14% |
| Stakeholder Communication & Lifecycle Management | 14% |
| Claude Models, Prompting & Context Engineering | 13% |
| Developer Productivity & Operational Enablement | 7% |
The capstone
The last lesson is a full system-design capstone: you defend one architecture across discovery, integration, evaluation, governance, and stakeholder lifecycle at once — the same cross-domain reasoning a real CCAR-P scenario question demands, instead of one domain at a time.
Labs, not slides
Every lesson links to a runnable lab in labs/ccar-p/lessons. The 9 exclusive lessons get their own new labs (RAG pipelines, observability dashboards, governance checklists, and more); the 16 shared lessons point back at the existing ccar-f/lessons labs instead of duplicating them — it's the same underlying decision, so it's the same code.
Free track, question bank in progress
The Study Track is free, full stop. CCAR-P's official question bank and timed mock exam are still in production on our end — the guide page with all the official exam details (fee, item count, prerequisites) is already live, and the paid simulado will land there once it's ready. In the meantime, the Study Track is the real preparation, not a placeholder for one.
Start the CCAR-P Study Track for free →
A note on where the structure came from
Like the CCAR-F track, the lesson sequence and pedagogical shape are adapted from rohitg00/ai-engineering-from-scratch (MIT licensed). All lesson content, quiz questions, and labs were written from scratch in our own voice — we only reused the curriculum's structure. Full attribution on our about page.