Description

For learners studying for Claude Certified Developer – Foundations
What you'll learn
Practice pack content
Domain 1 - Agents and Workflows
Domain 2 – Applications and Integration
Domain 3 – Claude Code
Domain 4 – Eval, Testing, and Debugging
Domain 5 – Model Selection and Optimization
Domain 6 – Prompt and Context Engineering
Domain 7 – Security and Safety
Domain 8 – Tools and MCPs
Final Exam Simulators
Exam Study Guide
Interactive Flashcards
Try 3 Claude Developer Foundations practice questions
Choose an answer, explore the explanation, and see why the other options are less suitable. Three short scenarios. No sign-in required.
Applications and IntegrationTopic · Claude API Mechanics
1. Restoring context in a follow-up request
A developer builds a troubleshooting assistant with direct Messages API calls and no session-management framework. The first turn describes a printer fault, but the next request contains only “Which of those checks should I try first?” and Claude lacks the earlier details. Which request change best restores the conversation context?
- Keep the same API key and model, and send only the new follow-up as the next user message.
- Send the earlier user turn, the assistant reply, and the follow-up in order in the messages array.
- Put the previous response ID inside the follow-up, and omit the earlier user and assistant turns.
- Keep earlier turns in application logs, and send only the follow-up with a larger max_tokens value.
Answer and full explanation
Correct answer: B — Include the prior exchange in the next Messages API request.
Why B is correct
The direct Messages API does not retain conversational state between requests. The application must supply the earlier turns that Claude needs for the next response. Here, both the printer description and the assistant’s suggested checks are relevant. Including those user and assistant messages before the new user message preserves their order and roles. This gives Claude the missing context without assuming that an API key or response identifier retrieves conversation history.
Why the other options are less suitable
- A — Reusing the API key and model: Keeping credentials and model selection consistent may be appropriate for the application. Neither choice supplies the omitted conversation to a stateless request.
- C — Sending only a response identifier: Response identifiers are useful for tracking individual API responses. Putting an identifier in the user’s text does not load the associated conversation into the request.
- D — Relying on logs and output capacity: Application logs can support troubleshooting, and a larger output limit can allow longer responses. Neither makes omitted log contents available to Claude.
Key takeaway: Manage conversation history in the application and include the relevant earlier turns in each direct Messages API call.
Read two more sample questions
Model Selection and OptimizationTopic · Cost and Token Management
2. A reusable prefix for independent questions
An API application answers independent questions about the same reference manual using explicit cache breakpoints. The manual exceeds the model’s minimum cacheable length, requests arrive within the configured cache lifetime, and other request settings remain unchanged. Which arrangement best reuses the manual’s cached prefix while allowing each question to change?
- Put the changing question before the manual, and place the cache breakpoint after the manual.
- Put the manual before the question, and insert a fresh request timestamp inside the manual block.
- Put the unchanged manual before the question, and place the cache breakpoint at the manual’s end.
- Put the manual before the question, and place the only cache breakpoint after the changing question.
Answer and full explanation
Correct answer: C — Cache the stable manual prefix before the changing question.
Why C is correct
Prompt caching reuses a matching prefix rather than an arbitrary passage anywhere in a request. Placing the unchanged manual first keeps it outside the changing question’s portion of the prompt. A breakpoint at the manual’s end establishes a cache entry for that reusable prefix. Later requests can reuse it while processing their new questions separately. The scenario already satisfies the length and lifetime requirements, so prefix placement is the decisive issue.
Why the other options are less suitable
- A — Putting variable content first: This order may be acceptable when caching is not a goal. Here, the changing question becomes part of the prefix before the manual, preventing that manual prefix from matching.
- B — Changing the manual block: A timestamp can be useful when the task needs time-specific information. Inserting a new value into the manual’s prefix changes the content that should remain reusable.
- D — Caching only after the variable suffix: A final breakpoint can work for a growing conversation that preserves earlier content. With independent changing questions and no manual-only cache entry, it does not establish the stable prefix needed here.
Key takeaway: Keep reusable content unchanged and place its explicit cache breakpoint before per-request variations.
Security and SafetyTopic · Claude Hooks
3. Blocking a production migration before execution
A team uses Claude Code with an MCP tool that accepts an environment parameter for running database migrations. Policy allows this tool in staging but forbids production calls, even when Claude proposes one. Which hook configuration can check that parameter and block the prohibited call before the tool executes?
- Use a PreToolUse command hook that checks the environment and returns a deny decision for production.
- Use a PostToolUse command hook that checks the environment and reports production calls afterward.
- Use a SessionStart command hook that reminds Claude to send migration requests only to staging.
- Use a Stop command hook that reviews the transcript and flags any migration targeting production.
Answer and full explanation
Correct answer: A — Deny prohibited calls with a PreToolUse command hook.
Why A is correct
PreToolUse runs after the tool arguments are created but before the tool call is processed. A command hook matched to the migration tool can inspect its environment parameter. For production, it can return hookSpecificOutput.permissionDecision set to "deny". That decision blocks the proposed call rather than relying on Claude to remember a written instruction. The migration service should also enforce its own authorization; the hook is an additional safeguard.
Why the other options are less suitable
- B — Checking after successful execution: PostToolUse can support logging or checks on completed operations. It runs after the tool succeeds, which is too late to prevent the production migration.
- C — Providing a session reminder: SessionStart can supply useful guidance when a session begins or resumes. A reminder does not inspect and deny each proposed migration call.
- D — Reviewing when Claude finishes: Stop hooks can check work before Claude finishes responding. Reviewing the transcript at that point cannot prevent a tool call that has already executed.
Key takeaway: Enforce tool-call restrictions before execution; reminders and retrospective checks serve different purposes.
Preview complete
You have tried all three samples
Review the explanations or try the samples again. This short preview covers only a small part of the exam objectives.
Your sample score resets when this page reloads.
Your practice pack includes
More ways to make each study session count
Focus on a topic, practise under a time limit, or return to the supporting resources before your next attempt.
Domain-based practice
Timed practice sets
Study guide
Flashcards
Requirements
- Anthropic recommends 1–5 years in software engineering and at least six months with Claude or comparable LLM systems, plus Python or TypeScript, REST API and CLI familiarity; these are recommendations, not mandatory experience requirements.
- Address gaps through foundational documentation and your own small Claude application, including API calls, a tool and output checks. Before booking the official exam, confirm eligibility through the Claude Partner Network registration process.
- Access this online practice product using an internet-connected computer, tablet or phone, a current web browser and an ExamsDigest account for the learning area.
Know the exam you’re studying for
These details describe Anthropic’s official Claude Certified Developer – Foundations exam.
What our students say
“I’d already used Claude’s API in a small project, but the ExamsDigest questions caught gaps I hadn’t noticed. The conversation-history questions were particularly helpful. Reading through the explanations showed me where I’d made assumptions instead of understanding how the requests worked.”
Ryan Mitchell
✓ Verified Review
“The explanations for the wrong answers were just as useful as the correct ones. They helped me work out what I needed to study next.”
Laura Bennett
✓ Verified Review
“Reading documentation was one thing; choosing between several plausible solutions was another. I used these questions alongside my Claude Developer Foundations notes, and the scenarios about caching and tool permissions gave me specific things to go back and test in my own project.”
Nathan Reed
✓ Verified Review
Your questions, answered.
Which exam and guide version do these questions target?
These questions target Claude Certified Developer – Foundations (CCDV-F), using the July 2026 exam guide, version 1.0. Match that code to your planned exam before purchasing, and check the product’s listed contents and access terms. The guide version identifies the blueprint used; it does not establish the number of questions, resources or official exam bookings included in a practice purchase.
Who is this Claude Developer practice for?
This practice is for developers working with Claude APIs, agents and application integrations, rather than people seeking only general chat or prompt-writing practice. Python or TypeScript, REST APIs, command-line tools and experience testing software will help you interpret the scenarios. Treat the recommended background as a study guide: revisit unfamiliar development concepts before repeatedly attempting the same questions.
How should I combine questions with hands-on development?
Combine questions with a small application you control, using non-sensitive test data. For example, maintain a two-turn Messages API conversation, add a read-only tool, and compare outputs against clear acceptance criteria. Change one implementation detail at a time and observe the result. Use this hands-on work alongside documentation; the practice questions themselves are not a live development environment.
What should I revisit after an incorrect answer?
Use each incorrect answer to identify the missing concept or overlooked requirement. Separate API integration mistakes from prompt, context, output-validation or permission issues, then revisit the relevant documentation and test a small example. Before retrying, explain why the correct choice fits and why your original choice does not. A higher practice score is not a guarantee of an official exam result.
Does CCDV-F use only single-answer questions?
The official exam contains multiple-choice and multiple-response items, with each question stating how many answers to select. The three-question preview uses only single-answer choices with explanations. It does not reproduce every official format, the full blueprint or a complete exam session. Check the paid product’s description for its actual question formats and study resources rather than assuming they match the preview.
Does completing the pack earn the Claude certification?
No. Completing this practice does not award an Anthropic credential or schedule an official exam. Pearson VUE directs candidates to register through Anthropic Partner Academy, and currently describes certification as open to Claude Partner Network organizations. Confirm your eligibility and follow the official booking process. Check the purchase terms for any voucher arrangements; practice access and official exam registration are separate.
30-day conditional refund policy
Eligible requests must be made within 30 calendar days, with no more than 20% of any included simulator completed. The purchase must not already be refunded or under a payment dispute or chargeback. Other provisions apply; mandatory consumer rights remain unaffected.
Useful links for your preparation



