Claude Certified Associate — Foundations

Sample Claude Certified Associate — Foundations practice questions

6 free sample questions from the practice bank, with the correct answer and the reasoning.

1. Question 1 · Prompting & Task Execution

   A team is drafting a reusable system prompt for a contract-review assistant. What is the correct order for its four components?

   • A.Output format → Role → Boundaries → Criteria
   • B.Role → Criteria → Boundaries → Output format Correct answer
   • C.Boundaries → Output format → Role → Criteria
   • D.Criteria → Boundaries → Role → Output format

   Why: Role first gives the model context before it reads the rules; output format comes last as the final shape constraint once everything else is established. Scrambling this order means later instructions have less context to work from.

2. Question 2 · Output Evaluation & Validation

   An internal HR chatbot occasionally gives answers that contradict the source policy document. What is required to measure this failure meaningfully?

   • A.p95/p99 response latency
   • B.Accuracy measured against a labeled dataset with known-correct answers Correct answer
   • C.Cost per request
   • D.Safety violation rate on adversarial prompts

   Why: Wrong or ungrounded answers are measured with accuracy against a labeled dataset. Latency, cost, and safety violation rate are each proxies for a different failure and won't catch factual incorrectness.

3. Question 3 · Product & Model Selection

   A 12-person marketing team runs the same weekly competitive analysis, re-pasting a 400-word instruction block into a new Claude.ai conversation every time. They have no engineering resources. What is the best fit?

   • A.The Claude API with a shared config file the team's code reads before each call
   • B.Claude.ai Projects, so the team can store shared, persistent instructions with no coding required Correct answer
   • C.Claude Code, since it supports reusable session instructions
   • D.A separate personal Claude.ai account per team member, each configured identically

   Why: Claude.ai Projects gives non-technical teams persistent, shared instructions with zero integration work — exactly this team's need. The API requires engineering the team doesn't have; separate accounts just recreate the copy-paste problem.

4. Question 4 · Workflow Integration & Solution Design

   During discovery for a new customer-support assistant, a stakeholder describes the capabilities they want but nothing else. Which discovery category is most often left unaddressed unless asked about explicitly?

   • A.What the system must do
   • B.What the system must NOT do — boundaries and cases that must route to a human Correct answer
   • C.What the system must cost
   • D.What the system must prove for audit purposes

   Why: Stakeholders describe desired capabilities readily but rarely volunteer prohibited actions or human-escalation boundaries on their own. This category has to be asked about explicitly, or it surfaces later — usually during compliance review, when it's expensive to fix.

5. Question 5 · Configuration & Knowledge Management

   A developer writes team coding standards into their personal `~/.claude/CLAUDE.md`. A new teammate clones the repository and Claude Code shows no sign of those standards. Why?

   • A.User-level CLAUDE.md is personal and never version-controlled — team-wide standards must be committed at project level Correct answer
   • B.CLAUDE.md files require an `@import` statement before they take effect
   • C.Only files under `.claude/rules/` are shared on clone
   • D.An admin must approve CLAUDE.md files before they distribute to teammates

   Why: `~/.claude/CLAUDE.md` is personal to the individual developer's machine and never travels with the repository. Team-wide conventions must live in the project-level `.claude/CLAUDE.md` (or root CLAUDE.md), which is committed and reaches every developer on checkout.

6. Question 6 · Governance, Risk & Responsible Use

   An employee says: 'Our company doesn't need an acceptable use policy for Claude — its built-in safety handles everything.' What is wrong with this statement?

   • A.Nothing — Claude's trained safety covers every business scenario automatically
   • B.Claude's trained safety covers broad, general-purpose harm reduction, but doesn't know an organization's specific data rules or authorization requirements — those must still be defined and enforced by the organization Correct answer
   • C.It's only wrong for companies handling health data or EU personal data
   • D.It's only wrong if the company operates outside the United States

   Why: Claude's built-in safety reduces obviously harmful output in general, but it has no knowledge of a specific company's data-handling rules, approval workflows, or acceptable-use boundaries. A request can pass Claude's safety checks entirely and still violate the organization's own policy.

What this practice exam covers

50 questions in 7 sections.

• Prompting & Task Execution
• Output Evaluation & Validation
• Product & Model Selection
• Workflow Integration & Solution Design
• Configuration & Knowledge Management
• Governance, Risk & Responsible Use
• Troubleshooting & Optimization

More practice for this exam

Claude Certified Associate: the 321-question bank

Questions about this practice exam

Are these the official Claude Certified Associate — Foundations exam questions?

No. They are practice questions written against the published exam domains. This site is not affiliated with or endorsed by Anthropic.

How many practice questions are there?

50 questions in 7 sections.

Do I need an account to practise?

No. You can take the practice questions without an account and your scores save in your browser. A free account adds the full answer review, gap analysis and a completion certificate.

Do the questions explain the answers?

Yes. Every question comes with an explanation of the correct answer.