Anthropic Claude Certified Developer - Foundations CCDV-F Exam Questions

Page: 1 / 14
Total 95 questions

Want more questions? Get Premium Access.
()

Question 1

A teammate has asked why your Claude application sometimes produces a response that includes the prompt text repeated back, and other times produces a response with the prompt text rephrased. They suspect a bug in the application's request construction.

How would you respond?

ATell the teammate that the variation depends on which Claude model serves the request, and recommend pinning the application to a single model version to make the output behavior consistent.

BExplain that LLMs generate output token by token, and variation in how prompt content appears in output is a property of generation, not a bug in request construction.

CConfirm that the variation is a bug in the application's request construction and start investigating which part of the application is producing the inconsistent prompt text.

DTell the teammate that the variation is caused by the application sending two different prompts on different runs and propose a code change that pins the system prompt to a single version.

Answer : B

Option B reflects the fundamental generative behavior of large language models. Claude does not retrieve a fixed response template for a given prompt. Instead, it generates the response incrementally, selecting subsequent tokens according to the probability distribution produced by the model. Consequently, semantically equivalent runs can differ in wording, ordering, whether prompt language is quoted, and whether it is paraphrased.

Anthropic's API documentation explicitly describes sampling parameters as controlling randomness in subsequent-token selection and also states that even configurations intended to minimize randomness do not make output completely deterministic. Current Claude model generations increasingly manage sampling internally, but the underlying principle remains: generative output is not guaranteed to be textually identical across equivalent calls.

Therefore, seeing the original prompt echoed in one response and paraphrased in another does not, by itself, demonstrate malformed request construction. A real request-construction problem should instead be established through traces showing different system instructions, messages, parameters, or context.

A, C, and D incorrectly infer specific implementation faults without supporting evidence.

The supplied question identifies B as the correct response. Relevant topics: SW Eng Foundations, token generation, sampling, nondeterminism, debugging, and model behavior.

===============

Question 2

The product team has asked you to choose a Claude model for a new feature. The team has provided functional requirements but has not specified performance, cost, or quality targets. The team's product manager says, "Use whatever model gives us the best results."

How would you respond?

AAsk the product team to specify quality, latency, and cost targets, then select the model whose tradeoffs best fit those targets.

BRun every Claude model on a representative sample and pick whichever scores best on a generic benchmark.

CChoose a mid-tier model and ship the feature, because mid-tier models work for most use cases without specified targets.

DChoose the largest, highest-capability Claude model, on the grounds that 'best results' is most likely to mean highest quality.

Answer : A

Option A is correct because model selection requires measurable non-functional requirements, not an undefined objective such as ''best.'' Anthropic's model-selection guidance explicitly frames the decision around capabilities, speed, and cost and recommends identifying those requirements before narrowing the candidate models.

''Quality'' must also be translated into task-specific success criteria---for example classification accuracy, extraction correctness, acceptable false-positive rate, or human preference. Latency may be constrained by an interactive SLA, and cost may be defined per request, per user, or per completed task. Once those thresholds exist, representative evaluations can compare candidate models using the actual prompts and data.

B uses a generic benchmark that may not correlate with the application's workload. C chooses a model tier without requirements or evidence. D optimizes only for raw capability and ignores latency and cost, potentially producing an unnecessarily expensive architecture.

Thus, requirements definition precedes technology selection. Relevant Study Guide topics: Understanding Reqs, functional versus non-functional requirements, quality thresholds, latency budgets, cost constraints, acceptance criteria, workload-specific evaluation, and model-selection tradeoffs.

===============

Question 3

Your Claude application has multi-step workflows where each step's output is needed only briefly before the agent moves on. The cumulative tool output is filling the context window with content that is no longer relevant.

How would you handle the accumulating tool output?

AApply tool output pruning to remove tool outputs that are no longer needed by later steps in the workflow.

BApply prompt caching to the accumulated tool outputs so the application does not re-pay for the older content on each subsequent step.

CSwitch to a smaller Claude model that processes context more efficiently and treat any quality loss as a tradeoff for the cost reduction.

DKeep every tool output in the context indefinitely so the agent has the full record of every step it has executed during the workflow.

Answer : A

Option A applies the correct context-engineering strategy: remove stale tool results once they no longer contribute useful information to subsequent reasoning. Agentic workflows frequently accumulate search results, file contents, API responses, and intermediate artifacts. Keeping all of them indefinitely consumes the finite context window, raises token cost, and can reduce model focus by introducing low-value information.

Anthropic specifically documents tool result clearing for this situation. Context Editing can remove older tool results when the conversation grows, while preserving recent interactions and optionally retaining tools whose results must remain available. Anthropic describes old tool outputs such as retrieved files or search results as candidates for clearing after Claude has processed them.

Prompt caching in B solves a different problem: it can lower cost and latency for repeated static prompt prefixes, but cached tokens still constitute context and therefore do not eliminate context-window pressure. C changes model capability without solving the architectural cause. D maximizes context pollution.

The correct architecture is therefore to preserve high-signal state while pruning ephemeral intermediate outputs. This aligns with Claude Developer coverage of context engineering, long-running agents, context-window management, tool-result clearing, and efficient agent state management. Anthropic's broader context-engineering guidance likewise emphasizes curating the smallest high-signal context necessary for successful inference.

===============

Question 4

You are designing an agent that handles a complex claim-processing workflow. Each claim moves through fact extraction, eligibility evaluation, and a decision step. The three subtasks have distinct success criteria, and some claims require iteration between fact extraction and eligibility evaluation before a decision can be reached.

Which agent pattern would you apply?

AA graph-based pattern that lets the agent move between subtasks based on the state of each claim, with each subtask evaluated against its own criteria.

BA streaming pattern that emits partial decisions as the agent processes each claim, refining the output until a final decision emerges from the stream.

CA single tool-use loop pattern that gives one agent access to all the tools needed for fact extraction, eligibility evaluation, and decision-making.

DA linear chain pattern that processes every claim through fact extraction, then eligibility evaluation, then decision, with no return paths between subtasks.

Answer : A

A is correct because the workflow is state-dependent, non-linear, and iterative. The source explicitly identifies the graph-based pattern as the intended architecture for this scenario. Each stage---fact extraction, eligibility evaluation, and decision-making---has its own completion criteria, and the process may need to move backward from eligibility evaluation to fact extraction when information is incomplete. A graph representation naturally models these conditional transitions and loops.

Anthropic's current orchestration guidance supports workflows containing branching, loops, filtering, staged execution, and state-dependent control flow, rather than forcing every task through one fixed sequence. Dynamic workflow orchestration can use explicit control logic so the next processing stage depends on current state and previous results.

B concerns progressive output delivery, not workflow-state transitions. C provides a generic agentic tool loop but does not explicitly model distinct states or transition criteria. D is unsuitable because it prohibits the required return path between extraction and eligibility evaluation.

Therefore, a graph-based architecture provides the necessary conditional routing, iteration, and stage-specific validation.

Relevant Claude Developer topics: Agent Patterns, graph workflows, state transitions, conditional branching, loops, stage-specific success criteria, and agent orchestration.

===============

Question 5

You are designing a Claude application that processes user-submitted text. Some of that text could include sensitive information such as account numbers or passwords that the application should not send to Claude.

How would you design the application?

ADefine the application boundary explicitly, identify what content can leave the boundary for Claude, and add filtering or redaction at the boundary.

BAdd a prompt instruction in the system prompt specifying the categories of sensitive information Claude should disregard when processing user-submitted text.

CLog all user-submitted text before it is sent to Claude and review the logs periodically to identify whether sensitive information is reaching the model.

DApply filtering at the boundary for the most commonly observed sensitive data patterns and expand coverage to additional patterns based on findings from production monitoring.

Answer : A

Option A is correct because sensitive-data protection must occur before prohibited information crosses the application's trust boundary. If account numbers, passwords, or other secrets must not be sent to Claude, asking Claude to ''ignore'' those values after transmission does not satisfy the requirement. The application must determine which data is permitted to leave its controlled environment and transform or reject prohibited content before constructing the Claude request.

This principle is consistent with Anthropic's defense-in-depth guidance. Anthropic recommends input screening, limiting Claude's access to sensitive information, and applying least privilege so the model receives only information necessary for the task.

B is fundamentally too late: the sensitive information has already been transmitted to the model context before the system instruction can influence behavior. C is worse because logging can create an additional store containing the sensitive information and only detects exposure after it occurs. D implements some boundary filtering but deliberately leaves coverage incomplete until production incidents reveal additional patterns.

Therefore, A requires the boundary and its permitted data flows to be defined first, then enforced comprehensively through filtering, redaction, tokenization, or rejection. Relevant Study Guide topics: application boundaries, data minimization, sensitive-data handling, filtering, redaction, least privilege, and privacy-preserving design.

===============

Question 6

Your Claude application's content policy specifies categories of content it should not produce under any circumstance. The application currently has no mechanism to enforce this policy, and content matching these categories is appearing in the application's output.

How would you enforce the content policy?

AEnhance the system prompt to contain explicit instructions for the categories to avoid, complete with examples of each category. Treat the strengthened prompt as the primary enforcement mechanism for the application's content policy across all responses.

BRemove the content policy entirely and let any output reach users during normal operation, accepting whatever content the application produces in response to incoming traffic.

CMove enforcement to users by asking them to report content policy violations after the violating content has already reached them in the application's responses.

DAdd deterministic output filtering that checks responses against the content policy before they reach users.

Answer : D

Option D is the strongest enforcement design because an unconditional content policy requires an application-level control between model generation and user delivery. Prompt instructions are valuable for steering Claude, but they are probabilistic controls and should not be treated as the sole enforcement mechanism when prohibited categories must never be exposed.

Anthropic's guardrail guidance recommends layered safeguards including screening, validation, monitoring, and filtering rather than relying exclusively on prompts. Its prompt-leak guidance specifically recommends output screening and post-processing, including deterministic techniques such as keyword matching, regular expressions, or other text-processing mechanisms where appropriate.

A improves the probability of policy compliance but cannot guarantee that every generated response will satisfy an externally defined application policy. B explicitly abandons the requirement. C detects violations only after exposure, which is unsuitable when the content must not reach users.

A production architecture can combine system instructions, structured classification, policy engines, deterministic rules, and model-based moderation, but the decisive requirement is enforcement before output delivery. Relevant Claude Developer topics are guardrails, output filtering, content moderation, deterministic enforcement, defense in depth, safe application boundaries, and production Claude application design.

===============

Question 7

A teammate is debugging a Claude application whose system prompt has grown to several hundred lines and now contains overlapping, contradictory, and obsolete instructions.

How would you advise the teammate?

AAdd more explicit instructions so the most recent rules dominate the model's interpretation of the prompt during each request.

BAudit the prompt for overlap, contradiction, and obsolete content, then refactor so each instruction is clear, current, and non-redundant.

CTighten only the contradictory rules first, treating the overlap and obsolete content as lower-priority work the team can address later.

DSplit the prompt across multiple system prompts so the model sees a smaller portion at any given time.

Answer : B

Option B is correct. A large prompt containing overlapping, contradictory, or obsolete instructions creates instruction ambiguity and increases the likelihood that Claude will apply an unintended rule, over-weight irrelevant guidance, or produce inconsistent behavior.

Anthropic's current prompting guidance reinforces this principle: Claude performs best when instructions are clear, explicit, and direct. Prompt design should minimize ambiguity and provide the model with precisely the context and behavioral requirements needed for the task. Refactoring should therefore remove obsolete requirements, consolidate duplicated instructions, resolve contradictions, and establish one authoritative formulation for each behavioral rule. The revised prompt should then be regression-tested against representative inputs.

A makes an already overloaded prompt larger and relies on instruction ordering instead of eliminating conflicts. C addresses only one defect while retaining redundant and obsolete material. D does not inherently solve contradictions and may fragment authoritative instructions across multiple locations.

Therefore, B addresses the root cause. Relevant Study Guide topics: system prompt design, prompt maintainability, instruction clarity, conflicting instructions, prompt refactoring, regression evaluation, and context quality.

===============