Free Quizzes

Free Claude Certified Architect - Foundations Practice Questions

20 exam-style Claude Certified Architect - Foundations questions with an explanation after every answer. No timer, no signup — about 15 minutes.

About the Claude Certified Architect - Foundations exam

The Claude Certified Architect - Foundations (CCAR-F) is a Foundations-level certification. It covers agentic architecture, tool and MCP design, Claude Code workflows, structured prompting, context management, and reliability. The exam has 60 questions and runs 120 minutes. You need a scaled score of 720 out of 1000 to pass.

The 20 practice questions on this page are original independent practice mapped to the public blueprint and official Claude learning material and balanced across the exam domains using the official exam guide weighting. They use the same scenario-based style as the real CCAR-F exam.

Related Claude Certified Architect - Foundations Practice Questions

Architecture & Workflow — CCAR-F Practice Question
Agentic Architecture and Orchestration
Tool Design & Schema — CCAR-F Practice Question
Tool Design and MCP Integration
Claude Code & Hierarchy — CCAR-F Practice Question
Claude Code Configuration and Workflows
Structured Output & Schema — CCAR-F Practice Question
Prompt Engineering and Structured Output
Context & Compaction — CCAR-F Practice Question
Context Management and Reliability
Multi Agent & Parallelization — CCAR-F Practice Question
Agentic Architecture and Orchestration

Frequently asked questions about the Claude Certified Architect - Foundations exam

Is Foundations an entry-level architecture exam?+

Foundations describes the breadth of the architecture credential, not a beginner chat-user exam. The public blueprint focuses on agent orchestration, tools and MCP, Claude Code workflows, structured output, context, and reliability.

What is the most important preparation pattern?+

Practice choosing between a flexible model instruction and a deterministic system control. For high-impact invariants, authorization, hooks, schemas, state machines, and approval boundaries usually provide stronger guarantees.

Are these real exam questions?+

No. CloudYeti publishes original independent scenarios only and does not reproduce protected exam content.

Sample Claude Certified Architect - Foundations questions

Three examples from the 20-question quiz. Answers stay hidden until you reveal them.

Sample Question 1Claude Code Configuration and Workflows

Claude Code opens a pull request after implementing a change. Which CI design best supports reliable agent-assisted delivery?

ASkip tests to preserve model context
BMerge automatically when the diff is non-empty
CRun deterministic tests, linters, security checks, and policy gates, then require appropriate review based on change risk
DAsk Claude whether its code is correct
Show answer
Answer: C — Run deterministic tests, linters, security checks, and policy gates, then require appropriate review based on change risk

External verification provides evidence independent of the agent's self-assessment. Review depth can scale with risk while CI enforces repeatable gates.

Sample Question 2Prompt Engineering and Structured Output

A prompt has accumulated forty examples and now behaves inconsistently when the examples conflict. What is the best next step?

AAdd forty more examples
BCurate a small representative non-conflicting set, move enforceable rules into explicit instructions or schemas, and rerun evals
CRemove all validation
DIncrease randomness
Show answer
Answer: B — Curate a small representative non-conflicting set, move enforceable rules into explicit instructions or schemas, and rerun evals

Examples should clarify behavior, not become an unmaintainable policy store. Explicit contracts and curated examples reduce contradictions and context cost.

Sample Question 3Agentic Architecture and Orchestration

A production agent can repeatedly search when evidence is weak. What architecture prevents unbounded loops while preserving useful autonomy?

AEnd after every first tool call
BRemove all observability to reduce latency
CLet it run until it feels finished
DTrack explicit stop reasons, tool-call budgets, elapsed time, repeated-call detection, and an escalation path
Show answer
Answer: D — Track explicit stop reasons, tool-call budgets, elapsed time, repeated-call detection, and an escalation path

Agent loops need explicit termination and budget policies. Repetition detection and escalation handle failure modes that model instructions alone may not stop.

Keep your progress

Sign in free to take these 20 questions in timed exam mode and keep your quiz history.