The Claude Certified Architect Foundations (CCA-F) exam: a complete guide
What the exam covers, the five domains and their weights, and a practical plan to prepare — whether you're new to building with Claude or already shipping agents.
What is the CCA-F exam?
The Claude Certified Architect — Foundations (CCA-F) is Anthropic's entry-level certification for people who design and build applications on Claude. It validates that you can make sound architectural decisions across the things that actually matter in production: orchestrating agents, configuring Claude Code, engineering reliable prompts, integrating tools and the Model Context Protocol (MCP), and keeping long-running systems reliable.
It is scenario-based rather than trivia: most questions put you in a realistic situation ("your agent loops on a failing tool call…", "you need structured output that downstream code can trust…") and ask for the best architectural choice. That means rote memorization helps far less than understanding why one approach beats another.
The five exam domains
The CCA-F blueprint is organized into five domains. Their relative weights tell you where to invest study time — Agentic Architecture & Orchestration alone is more than a quarter of the exam.
| Domain | Weight |
|---|---|
| Agentic Architecture & Orchestration (AAO) | 27% |
| Claude Code Configuration & Workflows (CCW) | 20% |
| Prompt Engineering & Structured Output (PESO) | 20% |
| Tool Design & MCP Integration (TDMI) | 18% |
| Context Management & Reliability (CMR) | 15% |
AAO · 27% Agentic Architecture & Orchestration
The biggest domain. When should a task be a single prompt versus a multi-step agent? How do you decompose work, hand off between sub-agents, add control flow and stopping conditions, and recover from a tool that fails or a loop that won't terminate? Expect questions on agent loops, planning vs. acting, parallel vs. sequential orchestration, and where deterministic code should wrap non-deterministic model calls.
CCW · 20% Claude Code Configuration & Workflows
Configuring Claude Code for real engineering work: project setup and CLAUDE.md conventions, permissions and safe tool use, hooks and slash commands, subagents, and choosing the right model and workflow for a task. Questions reward knowing which mechanism the harness actually provides versus what you'd have to build yourself.
PESO · 20% Prompt Engineering & Structured Output
Writing prompts that are precise and robust, and getting output a program can rely on: schemas and tool-call/JSON output, handling refusals and edge cases, few-shot vs. zero-shot, and reducing ambiguity. The emphasis is on structured output — validation, retries, and designing prompts whose results downstream code can trust.
TDMI · 18% Tool Design & MCP Integration
Designing tools Claude can use well — clear names, tight schemas, good descriptions — and integrating them via the Model Context Protocol (MCP). When to expose a capability as a tool, how to scope and secure it, error semantics, and how MCP servers fit into an application's architecture.
CMR · 15% Context Management & Reliability
Keeping systems reliable as context grows: managing the context window, prompt caching, summarization and retrieval, idempotency and retries, observability, and graceful degradation. The questions that separate a foundations-level architect from a beginner often live here.
How to prepare
- Weight your study to the blueprint. Spend the most time on AAO and the least on CMR — but don't skip any domain; the exam draws from all five.
- Practice with scenarios, not flashcards. Because the exam is scenario-based, the best preparation is answering scenario questions and reading the reasoning for why the right answer wins and the distractors lose.
- Build something small. Wire up a two-step agent with one tool and a stopping condition. The architectural trade-offs the exam tests become obvious once you've hit them yourself.
- Benchmark with full mock exams. Take timed mocks to find your weak domains, then drill those specifically before retesting.
Prepare with Brevyn
Brevyn is built specifically for the CCA-F: 200+ curated scenario questions anchored to all five domains, 10-question quizzes and full 60-question mock exams, and — for Pro — unlimited AI-generated practice that targets your weak domains, written by Claude Sonnet 4.6 and quality-checked by Claude Haiku 4.5. Free to start, no card required.
FAQ
When does the CCA-F exam launch?
Anthropic announced the Claude Certified Architect Foundations exam for 2026. Brevyn's questions are anchored to the published blueprint and production scenarios so you can prepare ahead of test day.
Is the exam multiple choice?
It is scenario-based: realistic situations with multiple-choice answers where the distractors are plausible-but-wrong architectural choices, not obvious throwaways.
How much should I study per domain?
Roughly match the weights — AAO 27%, CCW 20%, PESO 20%, TDMI 18%, CMR 15% — but cover all five, since every domain is tested.
Do I need to know Claude Code for the exam?
Yes — Claude Code configuration and workflows is a full 20% domain. Hands-on familiarity with project setup, permissions, hooks, subagents and model selection helps a lot.
Brevyn is an independent exam-preparation tool and is not affiliated with, endorsed by, or sponsored by Anthropic. "Claude" is a trademark of Anthropic. Exam details reflect publicly available information and may change; always check Anthropic's official certification pages for authoritative details.