Anthropic CCDV-F (Claude Certified Developer - Foundations) Practice Test 58

Question 570

The acceptance criteria for a production API change state that it must return an answer to a short user request through a backend service. Which implementation aligns with that criterion?

  1. Handle a refusal stop_reason safely
  2. Handle the max_tokens stop_reason
  3. Send tool_result content back to Claude
  4. Call the Messages API with explicit model and max_tokens
  5. Differentiate client and server tool execution

Explanation

Correct Answer

D:

A message request identifies the target model, a message list, and an output-token ceiling; this behavior directly supports the need to return an answer to a short user request through a backend service. The API is not a stateful chat session unless prior turns are supplied.

Incorrect Answers

A:

A refusal is a deliberate non-completion path that requires a defined application response; this applies to the different goal to avoid parsing safety refusals as valid JSON business records, not the requirement to return an answer to a short user request through a backend service; repeatedly increasing the token limit does not resolve a content refusal.

B:

When an application needs to recover a response that ended because its output budget was exhausted, the finish reason identifies truncation, so a caller may need a continuation or a revised token budget; the current capability must instead return an answer to a short user request through a backend service, while treating that response as a complete answer can hide missing fields.

C:

A result block tied to the tool call returns the external observation to the model for synthesis; the supported outcome is to produce a user-facing answer after a successful inventory lookup, whereas this scenario requires the ability to return an answer to a short user request through a backend service; a function's return value in server memory is invisible until included in the next request.

E:

The integration owns execution of client tools, while supported server tools are performed by the provider; despite being useful when a service must avoid assuming a custom local database function runs inside Anthropic infrastructure, it is not a solution to return an answer to a short user request through a backend service; tool names alone do not determine who executes the action.

Question 571

The owner of a model platform integration wants to verify that an optimization actually reuses an earlier prompt prefix. Which proposal provides the required mechanism?

  1. Set maximum step counts for agents
  2. Measure cache read tokens
  3. Keep volatile fields after reusable context
  4. Use model routing for simple versus hard cases

Explanation

Correct Answer

B:

Usage fields show how much eligible cached input was read rather than recomputed; this behavior directly supports the need to verify that an optimization actually reuses an earlier prompt prefix. Assuming identical UI prompts means a cache hit can conceal changed metadata or order.

Incorrect Answers

A:

A bound on iterations prevents repeated generation and calls from growing without limit; the supported outcome is to contain cumulative token and tool cost across a recursive research session, whereas this scenario requires the ability to verify that an optimization actually reuses an earlier prompt prefix; a per-message output cap alone does not control total multi-step spend.

C:

Stable instruction and reference sections should precede changing per-request details for effective prefix reuse; despite being useful when a service must prevent request timestamps from invalidating a costly static prefix, it is not a solution to verify that an optimization actually reuses an earlier prompt prefix; placing a changing value before the cache boundary undermines matching.

D:

For the alternate objective to spend high-capability tokens only when validation indicates complexity, conditional routing improves blended cost without discarding an escalation path; the feature here must verify that an optimization actually reuses an earlier prompt prefix, and blindly downgrading every call can increase retries and review expenses.

Question 572

A developer is reviewing alternatives for a prompt engineering feature that must reject a response that omits a mandated account identifier. Which option is a defensible fit?

  1. Use bounded repair or regeneration
  2. Check evidence alignment after extraction
  3. Separate a partial streaming fragment from a finished object
  4. Validate required schema fields

Explanation

Correct Answer

D:

Schema checks enforce field presence, types, and enumerations before the result enters a database; this behavior directly supports the need to reject a response that omits a mandated account identifier. Syntactically valid JSON may still be contract-invalid.

Incorrect Answers

A:

When an application needs to recover malformed data without entering an infinite prompt-retry loop, a limited correction attempt can use validation errors as feedback before escalating; the current capability must instead reject a response that omits a mandated account identifier, while unbounded retries can consume budget without resolving a defective input.

B:

Semantic verification compares fields to authoritative source spans; the supported outcome is to detect plausible dates or amounts that are absent from the provided document, whereas this scenario requires the ability to reject a response that omits a mandated account identifier; a high-confidence generated value is not proof that it was present.

C:

This approach serves the separate need to prevent the frontend from parsing an incomplete JSON chunk as a complete response because stream consumers must buffer and recognize completion before strict object validation; it does not establish the outcome to reject a response that omits a mandated account identifier, and an arbitrary content delta may end halfway through a field.

Question 573

A developer is reviewing alternatives for a tool integration feature that must require stricter approval for actions that change customer data. Which option is a defensible fit?

  1. Map tool_use IDs to results
  2. Restrict tool network credentials
  3. Define narrow tool input schemas
  4. Separate read tools from write tools

Explanation

Correct Answer

D:

Distinct interfaces simplify least-privilege grants and side-effect controls; this behavior directly supports the need to require stricter approval for actions that change customer data. A generic execute-anything function makes authorization difficult to enforce.

Incorrect Answers

A:

For the alternate objective to support multiple simultaneous lookups without mixing their responses, correlation IDs connect each observation to the exact invocation that requested it; the feature here must require stricter approval for actions that change customer data, and pairing results by arrival order is unsafe under concurrency.

B:

Minimal capabilities reduce risk when generated arguments or retrieved text are hostile; this applies to the different goal to stop a directory lookup function from also gaining unrestricted database write access, not the requirement to require stricter approval for actions that change customer data; better prompt wording does not remove excess infrastructure permissions.

C:

When an application needs to prevent ambiguous generated arguments from reaching an inventory service, typed parameter definitions constrain requests and enable strict validation before execution; the current capability must instead require stricter approval for actions that change customer data, while a prose-only command description cannot validate required field types.

Question 574

A Claude application application is approaching release, but the team still needs to keep an integration functional when a pinned model version approaches retirement. What is the most appropriate implementation?

  1. Set a maintenance and deprecation plan
  2. Instrument production observability
  3. Implement automated regression tests
  4. Define a staged release with rollback
  5. Separate development, staging, and production environments

Explanation

Correct Answer

A:

A dependency inventory and upgrade schedule create time for evaluation and migration; this behavior directly supports the need to keep an integration functional when a pinned model version approaches retirement. Indefinite reliance on a model identifier can become an outage risk.

Incorrect Answers

B:

For the alternate objective to identify rising error rates and latency after launch, logs, metrics, traces, and alert thresholds reveal operational failures at the right layer; the feature here must keep an integration functional when a pinned model version approaches retirement, and user feedback alone arrives too late for timely diagnosis.

C:

Repeatable cases compare changes against documented acceptance criteria; this applies to the different goal to catch schema and policy failures after a model or tool update, not the requirement to keep an integration functional when a pinned model version approaches retirement; a single successful demonstration provides little protection against regressions.

D:

When an application needs to deploy a prompt change without exposing every customer to an untested regression, gradual rollout and versioned rollback allow metrics to be compared while limiting impact; the current capability must instead keep an integration functional when a pinned model version approaches retirement, while editing the only production prompt in place makes reversions harder.

E:

Environment boundaries keep credentials, data, and writes isolated during testing; the supported outcome is to exercise an agent's database tool without touching live customer records, whereas this scenario requires the ability to keep an integration functional when a pinned model version approaches retirement; reusing production tokens in a local experiment defeats the separation.

Question 575

Two separate tickets in a model platform backlog require different controls. One needs to avoid selecting a model that lacks a required document or tool feature; the other needs to select the right model for an application with tool and network overhead. Which implementation pair is correct?

  1. First component: Prefer a Haiku-class model for lightweight routing. Second component: Prefer a high-capability Opus-class model.
  2. First component: Check modality and feature support. Second component: Measure end-to-end latency, not just token speed.
  3. First component: Check modality and feature support. Second component: Prefer a Haiku-class model for lightweight routing.
  4. First component: Prefer a high-capability Opus-class model. Second component: Measure end-to-end latency, not just token speed.

Explanation

Correct Answer

B:

The first proposal fits its assigned role because availability of context length, structured output, and modalities may depend on the selected model and version, while the second proposal fits its assigned role because total response time includes queuing, input processing, generation, and downstream integration; their respective acceptance criteria are to avoid selecting a model that lacks a required document or tool feature and to select the right model for an application with tool and network overhead.

Incorrect Answers

A:

The first proposal serves the different purpose to classify many short messages where low latency and unit cost dominate instead of the assigned first role because the lighter tier suits inexpensive, high-throughput tasks with modest reasoning demand, while the second proposal serves the different purpose to handle a rare complex architecture review where reasoning accuracy outweighs cost and latency instead of the assigned second role because the capability-oriented tier is suited to the hardest tasks after evaluation confirms an advantage; their respective acceptance criteria are to avoid selecting a model that lacks a required document or tool feature and to select the right model for an application with tool and network overhead.

C:

The first proposal fits its assigned role because availability of context length, structured output, and modalities may depend on the selected model and version, while the second proposal serves the different purpose to classify many short messages where low latency and unit cost dominate instead of the assigned second role because the lighter tier suits inexpensive, high-throughput tasks with modest reasoning demand; their respective acceptance criteria are to avoid selecting a model that lacks a required document or tool feature and to select the right model for an application with tool and network overhead.

D:

The first proposal serves the different purpose to handle a rare complex architecture review where reasoning accuracy outweighs cost and latency instead of the assigned first role because the capability-oriented tier is suited to the hardest tasks after evaluation confirms an advantage, while the second proposal fits its assigned role because total response time includes queuing, input processing, generation, and downstream integration; their respective acceptance criteria are to avoid selecting a model that lacks a required document or tool feature and to select the right model for an application with tool and network overhead.

Question 576

Two separate tickets in a credential management backlog require different controls. One needs to route borderline high-risk requests to a review process instead of automatic publication; the other needs to restore a safe service state after a new guardrail configuration causes widespread errors. Which implementation pair is correct?

  1. First component: Deploy content moderation and escalation. Second component: Build a documented rollback and incident process.
  2. First component: Use layered checks for unsafe requests. Second component: Build a documented rollback and incident process.
  3. First component: Deploy content moderation and escalation. Second component: Enforce least-privilege tool permissions.
  4. First component: Enforce least-privilege tool permissions. Second component: Use layered checks for unsafe requests.

Explanation

Correct Answer

A:

The first proposal fits its assigned role because a defined moderation workflow handles uncertain or sensitive material according to business policy, while the second proposal fits its assigned role because versioned release artifacts and response runbooks support containment and recovery; their respective acceptance criteria are to route borderline high-risk requests to a review process instead of automatic publication and to restore a safe service state after a new guardrail configuration causes widespread errors.

Incorrect Answers

B:

The first proposal serves the different purpose to detect harmful instructions at input, tool-call, and output boundaries instead of the assigned first role because independent controls catch failures that a single prompt-level rule may miss, while the second proposal fits its assigned role because versioned release artifacts and response runbooks support containment and recovery; their respective acceptance criteria are to route borderline high-risk requests to a review process instead of automatic publication and to restore a safe service state after a new guardrail configuration causes widespread errors.

C:

The first proposal fits its assigned role because a defined moderation workflow handles uncertain or sensitive material according to business policy, while the second proposal serves the different purpose to let a support assistant read shipment status without issuing refunds instead of the assigned second role because restricted scopes and read-only interfaces limit impact if a tool request is manipulated; their respective acceptance criteria are to route borderline high-risk requests to a review process instead of automatic publication and to restore a safe service state after a new guardrail configuration causes widespread errors.

D:

The first proposal serves the different purpose to let a support assistant read shipment status without issuing refunds instead of the assigned first role because restricted scopes and read-only interfaces limit impact if a tool request is manipulated, while the second proposal serves the different purpose to detect harmful instructions at input, tool-call, and output boundaries instead of the assigned second role because independent controls catch failures that a single prompt-level rule may miss; their respective acceptance criteria are to route borderline high-risk requests to a review process instead of automatic publication and to restore a safe service state after a new guardrail configuration causes widespread errors.

Question 577

A release plan for a Claude application platform has two acceptance checks: component one must deliver model-generated records to a strict financial reporting API; component two must ensure protected records are processed only in approved jurisdictions. Which proposal passes both?

  1. First component: Agree on output contracts with downstream systems. Second component: Set an explicit data residency constraint.
  2. First component: Define failure and recovery objectives. Second component: Set an explicit data residency constraint.
  3. First component: Specify human-review triggers. Second component: Define failure and recovery objectives.
  4. First component: Agree on output contracts with downstream systems. Second component: Specify human-review triggers.

Explanation

Correct Answer

A:

The first proposal fits its assigned role because typed fields and required identifiers enable validators and consumers to reject malformed results, while the second proposal fits its assigned role because documented residency requirements inform cloud placement, logging, storage, and vendor selection; their respective acceptance criteria are to deliver model-generated records to a strict financial reporting API and to ensure protected records are processed only in approved jurisdictions.

Incorrect Answers

B:

The first proposal serves the different purpose to keep a regulated triage service usable during an upstream API outage instead of the assigned first role because availability, recovery time, and fallback behavior must be specified as measurable operational needs, while the second proposal fits its assigned role because documented residency requirements inform cloud placement, logging, storage, and vendor selection; their respective acceptance criteria are to deliver model-generated records to a strict financial reporting API and to ensure protected records are processed only in approved jurisdictions.

C:

The first proposal serves the different purpose to require manual signoff when an AI-produced recommendation exceeds a financial threshold instead of the assigned first role because decision thresholds turn governance expectations into testable escalation rules, while the second proposal serves the different purpose to keep a regulated triage service usable during an upstream API outage instead of the assigned second role because availability, recovery time, and fallback behavior must be specified as measurable operational needs; their respective acceptance criteria are to deliver model-generated records to a strict financial reporting API and to ensure protected records are processed only in approved jurisdictions.

D:

The first proposal fits its assigned role because typed fields and required identifiers enable validators and consumers to reject malformed results, while the second proposal serves the different purpose to require manual signoff when an AI-produced recommendation exceeds a financial threshold instead of the assigned second role because decision thresholds turn governance expectations into testable escalation rules; their respective acceptance criteria are to deliver model-generated records to a strict financial reporting API and to ensure protected records are processed only in approved jurisdictions.

Question 578

An engineer extending an engineering productivity workflow has been asked to ensure a generated specification is validated before code generation begins. Which choice addresses the request most precisely?

  1. Use a deterministic workflow
  2. Use an agentic decision loop
  3. Use sequential chaining for dependent subtasks
  4. Adopt a supervisor with specialist workers

Explanation

Correct Answer

C:

Each stage consumes an approved artifact from the previous stage, preserving dependency order and stage-specific checks; this behavior directly supports the need to ensure a generated specification is validated before code generation begins. Parallel execution would allow downstream work to proceed on unverified assumptions.

Incorrect Answers

A:

A fixed pipeline keeps control flow in application code, making execution order reproducible and easy to audit; this applies to the different goal to execute an unchanged sequence of extraction, validation, and publishing steps, not the requirement to ensure a generated specification is validated before code generation begins; a free-running agent adds planning variability without helping when every step is already known.

B:

When an application needs to investigate an open-ended incident where the next query depends on earlier findings, a model-directed loop can choose subsequent actions based on tool observations until it reaches a stopping condition; the current capability must instead ensure a generated specification is validated before code generation begins, while a predetermined chain cannot reliably anticipate all branches of an exploratory diagnosis.

D:

This approach serves the separate need to coordinate separate billing, compliance, and engineering analyses under one final decision because a managing agent delegates bounded subtasks to specialists and reconciles their outputs against a common objective; it does not establish the outcome to ensure a generated specification is validated before code generation begins, and independent workers with no coordinator can return inconsistent recommendations.

Question 579

A document processing review asks the team to close two unrelated gaps without merging their execution paths. Service A must prevent API credentials from being exposed through shared repository history; service B must serve independent API calls without blocking a worker thread for each network wait. Which paired remediation works?

  1. First component: Keep secrets out of source control. Second component: Use asynchronous I/O for concurrent requests.
  2. First component: Use explicit HTTP method semantics. Second component: Separate integration tests from unit tests.
  3. First component: Separate integration tests from unit tests. Second component: Use asynchronous I/O for concurrent requests.
  4. First component: Keep secrets out of source control. Second component: Use explicit HTTP method semantics.

Explanation

Correct Answer

A:

The first proposal fits its assigned role because secret stores or environment injection keep production tokens out of committed code, while the second proposal fits its assigned role because async execution overlaps I/O-bound tasks while maintaining control over concurrency; their respective acceptance criteria are to prevent API credentials from being exposed through shared repository history and to serve independent API calls without blocking a worker thread for each network wait.

Incorrect Answers

B:

The first proposal serves the different purpose to prevent a supposedly read-only status lookup from changing a server-side record instead of the assigned first role because gET should retrieve resources while state changes use appropriate mutation methods and authorization checks, while the second proposal serves the different purpose to verify that a real staging tool and its API contract work together instead of the assigned second role because integration tests exercise boundaries and serialization that isolated mocks may miss; their respective acceptance criteria are to prevent API credentials from being exposed through shared repository history and to serve independent API calls without blocking a worker thread for each network wait.

C:

The first proposal serves the different purpose to verify that a real staging tool and its API contract work together instead of the assigned first role because integration tests exercise boundaries and serialization that isolated mocks may miss, while the second proposal fits its assigned role because async execution overlaps I/O-bound tasks while maintaining control over concurrency; their respective acceptance criteria are to prevent API credentials from being exposed through shared repository history and to serve independent API calls without blocking a worker thread for each network wait.

D:

The first proposal fits its assigned role because secret stores or environment injection keep production tokens out of committed code, while the second proposal serves the different purpose to prevent a supposedly read-only status lookup from changing a server-side record instead of the assigned second role because gET should retrieve resources while state changes use appropriate mutation methods and authorization checks; their respective acceptance criteria are to prevent API credentials from being exposed through shared repository history and to serve independent API calls without blocking a worker thread for each network wait.

Question 580

A technical lead requests a solution that can avoid assuming a custom local database function runs inside Anthropic infrastructure within a backend integration project. Which design should be chosen?

  1. Handle message_stop and final usage events
  2. Include an image content block
  3. Differentiate client and server tool execution
  4. Process tool_use responses in the application

Explanation

Correct Answer

C:

The integration owns execution of client tools, while supported server tools are performed by the provider; this behavior directly supports the need to avoid assuming a custom local database function runs inside Anthropic infrastructure. Tool names alone do not determine who executes the action.

Incorrect Answers

A:

For the alternate objective to record complete streamed output and token usage accurately, the final stream events signal completion and deliver accumulated metadata beyond individual text fragments; the feature here must avoid assuming a custom local database function runs inside Anthropic infrastructure, and stopping after the first content delta truncates responses.

B:

Image inputs are supplied as typed content with appropriate source data or references; this applies to the different goal to analyze a photographed shipping label with the supported multimodal Messages input, not the requirement to avoid assuming a custom local database function runs inside Anthropic infrastructure; describing a local image pathname in ordinary text does not transmit its pixels.

D:

Client tools execute in application code; the model emits structured requests but does not invoke external functions itself; the supported outcome is to actually retrieve a customer's order status requested by Claude, whereas this scenario requires the ability to avoid assuming a custom local database function runs inside Anthropic infrastructure; displaying the tool request as the final answer never performs the lookup.

Claude Certified Developer - Foundations (CCDV-F) Practice Tests