Anthropic CCDV-F (Claude Certified Developer - Foundations) Practice Test 33

Question 320

In a tool integration project, acceptance test A checks whether a component can use an already supported capability without hosting a duplicate custom service. Acceptance test B checks whether a different component can call a protected internal pricing API under the application's own credentials. Which plan satisfies both tests?

  1. First component: Choose a built-in provider tool. Second component: Create a custom client-side tool.
  2. First component: Choose a built-in provider tool. Second component: Combine tool access with human confirmation.
  3. First component: Use subagents for independent expertise. Second component: Create a custom client-side tool.
  4. First component: Combine tool access with human confirmation. Second component: Use subagents for independent expertise.

Explanation

Correct Answer

A:

The first proposal fits its assigned role because provider-managed execution can reduce implementation work when its scope matches the requirement, while the second proposal fits its assigned role because local execution gives the developer direct control over authentication, validation, and side effects; their respective acceptance criteria are to use an already supported capability without hosting a duplicate custom service and to call a protected internal pricing API under the application's own credentials.

Incorrect Answers

B:

The first proposal fits its assigned role because provider-managed execution can reduce implementation work when its scope matches the requirement, while the second proposal serves the different purpose to enable a customer service agent to prepare but not autonomously approve account closures instead of the assigned second role because tools supply actionable capabilities while a separate approval step governs sensitive execution; their respective acceptance criteria are to use an already supported capability without hosting a duplicate custom service and to call a protected internal pricing API under the application's own credentials.

C:

The first proposal serves the different purpose to delegate a security analysis while keeping the main coordinator's context concise instead of the assigned first role because specialist contexts reduce tool-output contamination and support focused parallel work, while the second proposal fits its assigned role because local execution gives the developer direct control over authentication, validation, and side effects; their respective acceptance criteria are to use an already supported capability without hosting a duplicate custom service and to call a protected internal pricing API under the application's own credentials.

D:

The first proposal serves the different purpose to enable a customer service agent to prepare but not autonomously approve account closures instead of the assigned first role because tools supply actionable capabilities while a separate approval step governs sensitive execution, while the second proposal serves the different purpose to delegate a security analysis while keeping the main coordinator's context concise instead of the assigned second role because specialist contexts reduce tool-output contamination and support focused parallel work; their respective acceptance criteria are to use an already supported capability without hosting a duplicate custom service and to call a protected internal pricing API under the application's own credentials.

Question 321

A Claude application solution has parallel requirements with separate owners. Workstream one must keep an interrupted invoice-creation request from generating duplicate invoices; workstream two must prevent a supposedly read-only status lookup from changing a server-side record. Which combination meets both?

  1. First component: Avoid retrying non-idempotent writes blindly. Second component: Use explicit HTTP method semantics.
  2. First component: Avoid retrying non-idempotent writes blindly. Second component: Validate webhook signatures.
  3. First component: Validate webhook signatures. Second component: Enforce concurrency limits.
  4. First component: Enforce concurrency limits. Second component: Use explicit HTTP method semantics.

Explanation

Correct Answer

A:

The first proposal fits its assigned role because retries of side-effecting actions need idempotency keys or transaction reconciliation, while the second proposal fits its assigned role because gET should retrieve resources while state changes use appropriate mutation methods and authorization checks; their respective acceptance criteria are to keep an interrupted invoice-creation request from generating duplicate invoices and to prevent a supposedly read-only status lookup from changing a server-side record.

Incorrect Answers

B:

The first proposal fits its assigned role because retries of side-effecting actions need idempotency keys or transaction reconciliation, while the second proposal serves the different purpose to reject forged callbacks from an external asynchronous job service instead of the assigned second role because cryptographic verification checks the sender and payload integrity before processing the event; their respective acceptance criteria are to keep an interrupted invoice-creation request from generating duplicate invoices and to prevent a supposedly read-only status lookup from changing a server-side record.

C:

The first proposal serves the different purpose to reject forged callbacks from an external asynchronous job service instead of the assigned first role because cryptographic verification checks the sender and payload integrity before processing the event, while the second proposal serves the different purpose to prevent bursts of document jobs from overwhelming rate limits and memory instead of the assigned second role because a semaphore or bounded queue keeps in-flight work within measured capacity; their respective acceptance criteria are to keep an interrupted invoice-creation request from generating duplicate invoices and to prevent a supposedly read-only status lookup from changing a server-side record.

D:

The first proposal serves the different purpose to prevent bursts of document jobs from overwhelming rate limits and memory instead of the assigned first role because a semaphore or bounded queue keeps in-flight work within measured capacity, while the second proposal fits its assigned role because gET should retrieve resources while state changes use appropriate mutation methods and authorization checks; their respective acceptance criteria are to keep an interrupted invoice-creation request from generating duplicate invoices and to prevent a supposedly read-only status lookup from changing a server-side record.

Question 322

The acceptance criteria for an engineering productivity change state that it must protect a coordinator's window from huge specialist tool outputs. Which implementation aligns with that criterion?

  1. Use an independent verifier step
  2. Use retrieval before generation
  3. Prefer a finite state machine for regulated steps
  4. Summarize long-running state with checkpoints
  5. Delegate narrow subtasks with a separate context

Explanation

Correct Answer

E:

Each subagent can work in a bounded context and return a concise, verified artifact; this behavior directly supports the need to protect a coordinator's window from huge specialist tool outputs. Copying every intermediate trace into the parent inflates context and cost.

Incorrect Answers

A:

A separate validation stage checks the artifact against explicit tests and constraints, limiting self-confirming errors; despite being useful when a service must detect mistakes in an agent-generated SQL migration before running it, it is not a solution to protect a coordinator's window from huge specialist tool outputs; asking the original agent whether it is correct is weaker than executable verification.

B:

For the alternate objective to answer questions from changing internal documentation with traceable source excerpts, retrieval grounds a model answer in current authoritative text rather than hoping parametric knowledge is current; the feature here must protect a coordinator's window from huge specialist tool outputs, and the retrieved text remains untrusted as instructions and needs relevance checks.

C:

Explicit states and transitions control who may advance a case independently of generated language; this applies to the different goal to enforce approval, execution, and confirmation order in a compliance workflow, not the requirement to protect a coordinator's window from huge specialist tool outputs; an unrestricted agent loop can skip expected governance stages.

D:

When an application needs to continue a support investigation after many rounds of log inspection, periodic compaction preserves decisions, open questions, and evidence references within the context budget; the current capability must instead protect a coordinator's window from huge specialist tool outputs, while blindly truncating oldest messages can remove critical constraints.

Question 323

An operations orchestration review asks the team to close two unrelated gaps without merging their execution paths. Service A must recover from a failed search while preserving the agent's chance to choose another path; service B must recover agent state when a container restarts during a lengthy customer case. Which paired remediation works?

  1. First component: Return a structured tool error to the model. Second component: Use a durable session store.
  2. First component: Use the Claude Agent SDK for an integrated tool-driven agent. Second component: Use a durable session store.
  3. First component: Return a structured tool error to the model. Second component: Keep user-specific state out of a shared agent process.
  4. First component: Keep user-specific state out of a shared agent process. Second component: Use the Claude Agent SDK for an integrated tool-driven agent.

Explanation

Correct Answer

A:

The first proposal fits its assigned role because tool-result errors surface the failure without impersonating a successful action, while the second proposal fits its assigned role because external persistence holds the conversation and workflow metadata needed to resume safely; their respective acceptance criteria are to recover from a failed search while preserving the agent's chance to choose another path and to recover agent state when a container restarts during a lengthy customer case.

Incorrect Answers

B:

The first proposal serves the different purpose to build a coding agent that needs managed tool execution, sessions, and event handling instead of the assigned first role because the SDK provides agent-loop primitives and integrations that remove much of the custom harness work, while the second proposal fits its assigned role because external persistence holds the conversation and workflow metadata needed to resume safely; their respective acceptance criteria are to recover from a failed search while preserving the agent's chance to choose another path and to recover agent state when a container restarts during a lengthy customer case.

C:

The first proposal fits its assigned role because tool-result errors surface the failure without impersonating a successful action, while the second proposal serves the different purpose to serve many tenants without mixing their tool outputs or prior messages instead of the assigned second role because per-request identity and session boundaries prevent cross-tenant context leakage; their respective acceptance criteria are to recover from a failed search while preserving the agent's chance to choose another path and to recover agent state when a container restarts during a lengthy customer case.

D:

The first proposal serves the different purpose to serve many tenants without mixing their tool outputs or prior messages instead of the assigned first role because per-request identity and session boundaries prevent cross-tenant context leakage, while the second proposal serves the different purpose to build a coding agent that needs managed tool execution, sessions, and event handling instead of the assigned second role because the SDK provides agent-loop primitives and integrations that remove much of the custom harness work; their respective acceptance criteria are to recover from a failed search while preserving the agent's chance to choose another path and to recover agent state when a container restarts during a lengthy customer case.

Question 324

A production-readiness check for an output validation component requires the service to solve a complex migration comparison that requires several distinct checks. Which approach passes this check?

  1. Use few-shot boundary cases
  2. Use a stepwise decomposition
  3. Specify output format in the prompt
  4. Delimit untrusted source material

Explanation

Correct Answer

B:

Breaking the work into verifiable stages lets the application check intermediate artifacts; this behavior directly supports the need to solve a complex migration comparison that requires several distinct checks. Simply increasing prompt length can make dependencies less clear.

Incorrect Answers

A:

Examples illustrating each side of a decision boundary communicate nuanced classification criteria; this applies to the different goal to distinguish near-duplicate support intents that a broad rule routinely confuses, not the requirement to solve a complex migration comparison that requires several distinct checks; many unrelated demonstrations can make the rule harder to infer.

C:

Explicit field descriptions guide generation toward the consumer's expected structure; the supported outcome is to obtain named fields in a consistent response layout for subsequent processing, whereas this scenario requires the ability to solve a complex migration comparison that requires several distinct checks; formatting guidance still requires machine validation at the boundary.

D:

This approach serves the separate need to let the model summarize a hostile email without obeying its embedded commands because clear source boundaries and explicit treatment as quoted evidence help reduce instruction confusion; it does not establish the outcome to solve a complex migration comparison that requires several distinct checks, and delimiters alone cannot replace permission checks on dangerous tools.

Question 325

A developer is reviewing alternatives for a document processing feature that must activate a new tool execution path for only internal testers. Which option is a defensible fit?

  1. Apply dependency injection for external clients
  2. Separate integration tests from unit tests
  3. Use feature flags for risky behavior changes
  4. Use explicit HTTP method semantics

Explanation

Correct Answer

C:

Runtime controls can segment rollout and permit quick disabling without a code rebuild; this behavior directly supports the need to activate a new tool execution path for only internal testers. Hard-coded unconditional activation exposes every tenant simultaneously.

Incorrect Answers

A:

This approach serves the separate need to test application logic without issuing real model calls or payments because injected interfaces allow deterministic stubs and reduce coupling to vendors; it does not establish the outcome to activate a new tool execution path for only internal testers, and hard-coded global clients make behavior harder to isolate and simulate.

B:

Integration tests exercise boundaries and serialization that isolated mocks may miss; despite being useful when a service must verify that a real staging tool and its API contract work together, it is not a solution to activate a new tool execution path for only internal testers; unit tests can pass while authentication or payload mapping is broken.

D:

GET should retrieve resources while state changes use appropriate mutation methods and authorization checks; this applies to the different goal to prevent a supposedly read-only status lookup from changing a server-side record, not the requirement to activate a new tool execution path for only internal testers; putting write behavior behind a read endpoint undermines predictable retry and caching semantics.

Question 326

A model platform review requires two independent, explicitly defined outcomes: the first workstream must implement an interactive collaboration channel requiring messages in both directions; a second must reduce setup overhead when a service sends many short requests. Which TWO actions directly provide these capabilities? Choose TWO.

  1. Use the official SDK for standard requests
  2. Choose WebSockets for bidirectional realtime exchange
  3. Use connection pooling for repeated HTTP calls
  4. Choose SSE for one-way streamed responses
  5. Use a queue for burst absorption

Explanation

Correct Answers

B:

Persistent two-way channels fit interactive event exchange where the client also sends messages during a session; this directly supports task 1 to implement an interactive collaboration channel requiring messages in both directions, not the separate task or tasks to reduce setup overhead when a service sends many short requests; standard response streaming alone is not a general bidirectional messaging protocol.

C:

Reusable clients avoid unnecessary connection establishment under compatible networking conditions; this directly supports task 2 to reduce setup overhead when a service sends many short requests, not the separate task or tasks to implement an interactive collaboration channel requiring messages in both directions; creating a fresh client per request can increase latency and resource churn.

Incorrect Answers

A:

Maintained client libraries wrap the HTTP API and expose structured response objects; this implements the different outcome to simplify authentication, serialization, and response typing in a Python service rather than the required tasks to reduce setup overhead when a service sends many short requests and to implement an interactive collaboration channel requiring messages in both directions, and SDK convenience does not eliminate the need for correct retry and error handling, so it is not a direct solution to any of the assigned workstreams.

D:

Server-sent events provide ordered server-to-client updates over an HTTP stream; this implements the different outcome to deliver tokens to a browser without maintaining a bidirectional command channel rather than the required tasks to reduce setup overhead when a service sends many short requests and to implement an interactive collaboration channel requiring messages in both directions, and a WebSocket is not required solely to push a response as it arrives, so it is not a direct solution to any of the assigned workstreams.

E:

Queueing decouples producers from consumer throughput and enables controlled retries; this implements the different outcome to smooth sudden ingestion spikes before workers call a capacity-limited model API rather than the required tasks to reduce setup overhead when a service sends many short requests and to implement an interactive collaboration channel requiring messages in both directions, and increasing frontend timeouts does not eliminate backend capacity limits, so it is not a direct solution to any of the assigned workstreams.

Question 327

Two separate tickets in a data pipeline backlog require different controls. One needs to keep an integration functional when a pinned model version approaches retirement; the other needs to uncover prompt-injection paths through retrieved webpages and connected tools. Which implementation pair is correct?

  1. First component: Assign an owner to every integration. Second component: Instrument production observability.
  2. First component: Set a maintenance and deprecation plan. Second component: Assign an owner to every integration.
  3. First component: Set a maintenance and deprecation plan. Second component: Perform a threat review before go-live.
  4. First component: Instrument production observability. Second component: Perform a threat review before go-live.

Explanation

Correct Answer

C:

The first proposal fits its assigned role because a dependency inventory and upgrade schedule create time for evaluation and migration, while the second proposal fits its assigned role because reviewing trust boundaries and permissions before deployment reveals attack routes unit tests may miss; their respective acceptance criteria are to keep an integration functional when a pinned model version approaches retirement and to uncover prompt-injection paths through retrieved webpages and connected tools.

Incorrect Answers

A:

The first proposal serves the different purpose to prevent abandoned connectors from retaining access after an application is transferred instead of the assigned first role because ownership enables rotation, monitoring, change approval, and retirement decisions, while the second proposal serves the different purpose to identify rising error rates and latency after launch instead of the assigned second role because logs, metrics, traces, and alert thresholds reveal operational failures at the right layer; their respective acceptance criteria are to keep an integration functional when a pinned model version approaches retirement and to uncover prompt-injection paths through retrieved webpages and connected tools.

B:

The first proposal fits its assigned role because a dependency inventory and upgrade schedule create time for evaluation and migration, while the second proposal serves the different purpose to prevent abandoned connectors from retaining access after an application is transferred instead of the assigned second role because ownership enables rotation, monitoring, change approval, and retirement decisions; their respective acceptance criteria are to keep an integration functional when a pinned model version approaches retirement and to uncover prompt-injection paths through retrieved webpages and connected tools.

D:

The first proposal serves the different purpose to identify rising error rates and latency after launch instead of the assigned first role because logs, metrics, traces, and alert thresholds reveal operational failures at the right layer, while the second proposal fits its assigned role because reviewing trust boundaries and permissions before deployment reveals attack routes unit tests may miss; their respective acceptance criteria are to keep an integration functional when a pinned model version approaches retirement and to uncover prompt-injection paths through retrieved webpages and connected tools.

Question 328

Before a latency optimization release, three separately owned tasks must be covered: one service must distinguish borderline policy labels with examples of both allowed and disallowed cases; another must brainstorm diverse naming candidates without demanding identical wording on every run; and a third must design tests that tolerate benign wording variation while detecting semantic failures. Which THREE choices directly address the stated tasks? Choose THREE.

  1. Use contrasting few-shot examples
  2. Use low-temperature generation for repeatable extraction
  3. Apply zero-shot instructions for a straightforward task
  4. Treat model outputs as stochastic
  5. Use one representative example
  6. Use varied sampling only for ideation

Explanation

Correct Answers

A:

Multiple representative cases help define a decision boundary and reduce ambiguity; this directly supports task 1 to distinguish borderline policy labels with examples of both allowed and disallowed cases, not the separate task or tasks to brainstorm diverse naming candidates without demanding identical wording on every run and to design tests that tolerate benign wording variation while detecting semantic failures; repeating near-identical samples does not cover edge cases.

D:

Non-determinism requires evaluation by criteria and repeated examples rather than exact-string equality everywhere; this directly supports task 3 to design tests that tolerate benign wording variation while detecting semantic failures, not the separate task or tasks to distinguish borderline policy labels with examples of both allowed and disallowed cases and to brainstorm diverse naming candidates without demanding identical wording on every run; a single run is insufficient to prove robust model behavior.

F:

Greater sampling diversity can produce a wider range of suggestions for exploratory work; this directly supports task 2 to brainstorm diverse naming candidates without demanding identical wording on every run, not the separate task or tasks to distinguish borderline policy labels with examples of both allowed and disallowed cases and to design tests that tolerate benign wording variation while detecting semantic failures; it is not appropriate as the sole correctness control for structured production data.

Incorrect Answers

B:

Lower randomness often improves consistency for narrow, constrained tasks; this implements the different outcome to reduce variation when a service maps documents into consistent categories rather than the required tasks to brainstorm diverse naming candidates without demanding identical wording on every run and to distinguish borderline policy labels with examples of both allowed and disallowed cases and to design tests that tolerate benign wording variation while detecting semantic failures, and it does not guarantee identical output or replace schema validation, so it is not a direct solution to any of the assigned workstreams.

C:

Direct instructions can be sufficient when the task is simple and well specified; this implements the different outcome to classify a familiar support message where labels and criteria are already clear rather than the required tasks to brainstorm diverse naming candidates without demanding identical wording on every run and to distinguish borderline policy labels with examples of both allowed and disallowed cases and to design tests that tolerate benign wording variation while detecting semantic failures, and adding examples unnecessarily consumes context and can introduce biases, so it is not a direct solution to any of the assigned workstreams.

E:

A single demonstration can anchor the expected shape and tone of a generated response; this implements the different outcome to clarify a specific formatting convention that a plain instruction was not conveying rather than the required tasks to brainstorm diverse naming candidates without demanding identical wording on every run and to distinguish borderline policy labels with examples of both allowed and disallowed cases and to design tests that tolerate benign wording variation while detecting semantic failures, and an example is not a formal guarantee that every response will conform, so it is not a direct solution to any of the assigned workstreams.

Question 329

During acceptance testing for an integration quality service, one requirement is to diagnose a request that fails consistently before generating any model output. What should the developer implement?

  1. Investigate a 401 authentication failure
  2. Validate data before blaming model quality
  3. Investigate a 400 invalid_request error
  4. Investigate a 403 authorization failure
  5. Interpret max_tokens as a truncation cause

Explanation

Correct Answer

C:

Validation failures point to malformed inputs or unsupported parameters that must be corrected; this behavior directly supports the need to diagnose a request that fails consistently before generating any model output. Backoff does not fix a structurally invalid payload.

Incorrect Answers

A:

Authentication errors require checking credential validity and configuration; this applies to the different goal to restore an integration after an API key is revoked, not the requirement to diagnose a request that fails consistently before generating any model output; a larger output-token budget cannot authenticate the caller.

B:

When an application needs to diagnose a generated summary that contradicts the latest database balance, source payloads, retrieval filters, and timestamps must be checked before attributing a wrong answer to inference; the current capability must instead diagnose a request that fails consistently before generating any model output, while prompt changes cannot repair a stale upstream ledger response.

D:

This approach serves the separate need to determine why a valid service identity is denied a protected capability because authorization failures concern permission or policy after identity is established; it does not establish the outcome to diagnose a request that fails consistently before generating any model output, and rotating an otherwise valid key will not necessarily grant missing access.

E:

The stop reason indicates the generation budget was exhausted rather than the task being completed; despite being useful when a service must explain why a generated answer ends before a required final field, it is not a solution to diagnose a request that fails consistently before generating any model output; a valid parse of the partial prefix does not make the report complete.

Question 330

The current design of a privacy controls pipeline does not adequately block a generated database query that targets another tenant's records. Which improvement resolves the gap?

  1. Add an explicit export approval check
  2. Validate tool arguments server-side
  3. Minimize personally identifiable information
  4. Use output escaping to prevent XSS

Explanation

Correct Answer

B:

Backend authorization checks must bind the requested operation to the caller's entitlements; this behavior directly supports the need to block a generated database query that targets another tenant's records. Well-formed arguments are not proof of permission.

Incorrect Answers

A:

When an application needs to prevent unreviewed bulk downloads triggered by an agent's tool plan, a separate authorization or human approval step can gate high-impact operations; the current capability must instead block a generated database query that targets another tenant's records, while the model's confidence in its own plan does not grant export permission.

C:

This approach serves the separate need to avoid transmitting unnecessary customer identifiers with a summarization request because redaction and selective fields reduce exposure before data enters model or logging paths; it does not establish the outcome to block a generated database query that targets another tenant's records, and logging every full prompt can defeat otherwise careful minimization.

D:

Context-aware encoding prevents text from being interpreted as active HTML or JavaScript; despite being useful when a service must display generated customer notes safely in a browser, it is not a solution to block a generated database query that targets another tenant's records; an instruction to the model not to write scripts is not a rendering control.

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