Anthropic CCDV-F (Claude Certified Developer - Foundations) Practice Test 20

Question 190

A web worker handles dozens of independent Claude calls without tying up one thread per waiting network request. Which solution matches the actual constraint?

  1. Use a work queue for long offline jobs
  2. Use async I/O for many concurrent inference requests
  3. Use POST for a Messages generation request
  4. Cancel timed-out tasks cleanly

Explanation

Correct Answer

B:

Asynchronous clients yield during I/O and resume when responses arrive; the implementation must still recognize that an event loop does not make a single model inference computationally faster.

Incorrect Answers

A:

A persistent queue decouples submission from workers and supports retry state; this addresses a separate concern, not the fact that asynchronous clients yield during I/O and resume when responses arrive; keeping a web request open indefinitely is fragile under client disconnects.

C:

The endpoint creates a new inference result from a JSON request and is invoked as an HTTP POST; the missing capability is more specific: asynchronous clients yield during I/O and resume when responses arrive; treating it as a cacheable read-only GET would not express the required request body.

D:

Cancellation and timeout handling release resources and prevent orphaned work where supported; by comparison, the intended solution relies on this: asynchronous clients yield during I/O and resume when responses arrive; merely hiding the client spinner does not stop backend consumption.

Question 191

A broad implementation task requires separate dependency inspection without adding every raw result to the parent context. Which implementation mechanism is needed?

  1. Use a sandbox for model-requested code execution
  2. Use a hook for mandatory pre-execution validation
  3. Use a custom command for a repeated invocation
  4. Use a subagent for context-isolated auditing

Explanation

Correct Answer

D:

A delegated worker returns focused findings while keeping its verbose exploration isolated—with this limitation: separate context is not a substitute for least-privilege tool access.

Incorrect Answers

A:

Process and network isolation contain potential side effects of generated programs; the needed capability is different—a delegated worker returns focused findings while keeping its verbose exploration isolated—and a model's internal safety judgment is not an OS permission boundary.

B:

A lifecycle hook evaluates the proposed action outside the model and can block it deterministically, whereas the requirement calls for another behavior: a delegated worker returns focused findings while keeping its verbose exploration isolated; natural-language checklists are weaker than enforced execution controls.

C:

A command encapsulates a recognizable workflow entry point for interactive coding sessions; however, the critical mechanism here is that a delegated worker returns focused findings while keeping its verbose exploration isolated, and it does not create a new callable external service by itself.

Question 192

A contract scan cannot locate a signature date but the output field is required by the application. Which mechanism should the team prioritize?

  1. Return an explicit unknown for missing evidence
  2. Escalate low-confidence high-impact cases
  3. Escape text before inserting into HTML
  4. Cross-check numerical fields with deterministic logic

Explanation

Correct Answer

A:

A nullable or unknown state avoids inventing a plausible but unsupported date; the important constraint is that guessing a value would convert absence of evidence into false data.

Incorrect Answers

B:

Risk-based human review catches consequential uncertain outputs before automatic action; the requirement instead turns on this behavior: a nullable or unknown state avoids inventing a plausible but unsupported date; automatic acceptance based on polished prose could cause compliance failures.

C:

Output encoding ensures untrusted characters display as text rather than executable DOM content; this cannot substitute for the behavior that a nullable or unknown state avoids inventing a plausible but unsupported date, because JSON validation is not a browser security measure.

D:

Code verifies arithmetic consistency after the model proposes the values; in contrast, a nullable or unknown state avoids inventing a plausible but unsupported date, while model confidence does not establish accounting correctness.

Question 193

A coding assistant should follow a reusable security-review procedure, query a read-only repository tool, and block shell commands that violate policy before execution. Which TWO combinations provide both instruction-plus-data and deterministic enforcement? Choose TWO.

  1. Combine MCP resources with a summarization prompt
  2. Use a subagent for context-isolated auditing
  3. Combine a guarded custom tool with a human approval step
  4. Combine a Skill with a read-only tool
  5. Combine a deterministic hook with a coding agent

Explanation

Correct Answers

D:

The Skill supplies the workflow while the tool supplies authorized external facts; notably, instruction packages and executable integrations solve different needs.

E:

A hook enforces the external check while the model handles the flexible development task; its operational boundary is that without correct hook wiring and tool permissions, guidance alone is insufficient.

Incorrect Answers

A:

Resources supply source data while a prompt governs how it should be analyzed, whereas the requirement calls for another behavior: the Skill supplies the workflow while the tool supplies authorized external facts; neither component independently guarantees the summary is factually faithful.

B:

A delegated worker returns focused findings while keeping its verbose exploration isolated; however, the critical mechanism here is that the Skill supplies the workflow while the tool supplies authorized external facts, and separate context is not a substitute for least-privilege tool access.

C:

The tool executes a constrained operation after an enforced business approval gate; the requirement instead turns on this behavior: the Skill supplies the workflow while the tool supplies authorized external facts; model-generated consent text is not an auditable approval event; the requested lookup is read-only and must be coupled to a deterministic command-screening step instead.

Question 194

A client needs to initiate a controlled customer lookup through a reusable integration. Which development approach should be selected?

  1. Expose an MCP tool for an executable operation
  2. Advertise tool metadata clearly
  3. Expose a custom API adapter outside MCP when local-only
  4. Expose an MCP resource for read-oriented context

Explanation

Correct Answer

A:

An MCP tool advertises callable functionality with input schemas and returns execution results; in production, tool calls still need explicit runtime permissions and backend validation.

Incorrect Answers

B:

The server supplies discoverable names, descriptions, and input schemas for the tool set; the needed capability is different—an MCP tool advertises callable functionality with input schemas and returns execution results—and misleading descriptions can cause inappropriate model tool selection.

C:

A direct adapter may be simpler than running and maintaining a separate protocol server, whereas the requirement calls for another behavior: an MCP tool advertises callable functionality with input schemas and returns execution results; choosing MCP solely for naming consistency introduces unnecessary lifecycle work.

D:

Resources represent accessible data that a client can retrieve and supply as context; however, the critical mechanism here is that an MCP tool advertises callable functionality with input schemas and returns execution results, and a resource's text cannot authorize additional tool actions by itself.

Question 195

Application teams need a stable classify_invoice function rather than repeated low-level prompt construction. What change is best aligned with the goal?

  1. Use a wrapper service for shared organizational controls
  2. Use an official SDK for conventional API calls
  3. Use HTTP request-response for discrete inference
  4. Use a typed abstraction for business-specific inference

Explanation

Correct Answer

D:

A domain wrapper hides model-request details behind validated inputs and outputs; its operational boundary is that abstractions should expose meaningful errors instead of swallowing model uncertainty.

Incorrect Answers

A:

A centralized proxy can enforce cross-cutting organizational controls before forwarding requests; however, the critical mechanism here is that a domain wrapper hides model-request details behind validated inputs and outputs, and the proxy also becomes a reliability and credential-protection responsibility.

B:

A supported client library wraps the underlying service API and common network details; the requirement instead turns on this behavior: a domain wrapper hides model-request details behind validated inputs and outputs; the SDK does not decide business-level retries or permissions automatically.

C:

Standard HTTPS requests fit isolated generation operations without long-lived bidirectional connections; this cannot substitute for the behavior that a domain wrapper hides model-request details behind validated inputs and outputs, because opening a persistent socket adds state and lifecycle complexity where it is unnecessary.

Question 196

An engineering team needs a reliable event record after selected tool actions complete. Which engineering practice is the strongest match?

  1. Set a maximum tool iteration count
  2. Use a pre-tool hook for prohibited actions
  3. Use a post-tool hook for outcome auditing
  4. Use a mandatory human approval mechanism

Explanation

Correct Answer

C:

A lifecycle hook can capture sanitized execution results and support monitoring or follow-up checks—with this limitation: a post-action hook cannot retroactively prevent a harmful command that already ran.

Incorrect Answers

A:

A harness enforces a fixed ceiling on action cycles and reports incomplete execution safely, whereas the requirement calls for another behavior: a lifecycle hook can capture sanitized execution results and support monitoring or follow-up checks; model self-assessment is too probabilistic to serve as the only loop brake.

B:

A pre-execution hook can inspect and block the operation before it reaches the shell; however, the critical mechanism here is that a lifecycle hook can capture sanitized execution results and support monitoring or follow-up checks, and the hook runs outside the model's judgment, but its own policy must be correctly implemented.

D:

An enforced approval boundary stops execution until confirmation is provided; this cannot substitute for the behavior that a lifecycle hook can capture sanitized execution results and support monitoring or follow-up checks, because a model's promise to ask first is not itself a technical approval gate.

Question 197

A help desk assistant must acknowledge submitted cases while the model provider is temporarily unavailable. Which development approach should be selected?

  1. Mock model responses for deterministic unit tests
  2. Use a safe degraded response on dependency outage
  3. Use a circuit breaker for cascading 5xx failures
  4. Escalate repeated unknown errors with trace context

Explanation

Correct Answer

B:

A fallback can queue the request or return a limited verified response without fabricating model results; retrying indefinitely inside the user's HTTP connection harms availability.

Incorrect Answers

A:

A mock response supplies controlled payloads so application logic can be tested repeatably; by comparison, the intended solution relies on this: a fallback can queue the request or return a limited verified response without fabricating model results; it does not prove a real model will produce the same semantic output.

C:

A circuit breaker temporarily blocks repeated failing calls while permitting controlled recovery probes, whereas the requirement calls for another behavior: a fallback can queue the request or return a limited verified response without fabricating model results; it should not be used to hide permanent programming errors.

D:

Capturing sanitized request IDs and relevant version metadata makes specialist diagnosis possible; however, the critical mechanism here is that a fallback can queue the request or return a limited verified response without fabricating model results, and silently dropping the incident gives no route to correction.

Question 198

A multi-tenant service must prevent one organization from consuming the entire shared inference allowance. Which development approach should be selected?

  1. Compute cost from input and output token usage
  2. Alert on unusual token spikes
  3. Forecast p95 cost rather than just average
  4. Set a per-customer token budget

Explanation

Correct Answer

D:

Usage attribution and quota enforcement limit spend by tenant and billing period, subject to this requirement: a global API key does not automatically impose tenant-level financial fairness.

Incorrect Answers

A:

Per-model input and output token rates combine with measured usage for accurate costing; the requirement instead turns on this behavior: usage attribution and quota enforcement limit spend by tenant and billing period; number of messages alone is not an adequate unit of resource consumption.

B:

Cost monitoring can detect and halt anomalous demand before it causes significant overspend; this cannot substitute for the behavior that usage attribution and quota enforcement limit spend by tenant and billing period, because monthly billing review alone detects runaway costs too late.

C:

Distribution-aware forecasting accounts for long-tail tasks and tool iterations; in contrast, usage attribution and quota enforcement limit spend by tenant and billing period, while median request cost can hide budget risk from heavy cases.

Question 199

A support product has rare expensive investigations that dominate peak operational spending. What would best address the stated failure mode?

  1. Set a per-customer token budget
  2. Tie cost optimization to measured quality
  3. Forecast p95 cost rather than just average
  4. Compute cost from input and output token usage

Explanation

Correct Answer

C:

Distribution-aware forecasting accounts for long-tail tasks and tool iterations—with this limitation: median request cost can hide budget risk from heavy cases.

Incorrect Answers

A:

Usage attribution and quota enforcement limit spend by tenant and billing period, whereas the requirement calls for another behavior: distribution-aware forecasting accounts for long-tail tasks and tool iterations; a global API key does not automatically impose tenant-level financial fairness.

B:

Total cost analysis includes downstream review and failure impact, not just provider charges; however, the critical mechanism here is that distribution-aware forecasting accounts for long-tail tasks and tool iterations, and a cheaper inference call is not a net saving if error handling multiplies workload.

D:

Per-model input and output token rates combine with measured usage for accurate costing; this cannot substitute for the behavior that distribution-aware forecasting accounts for long-tail tasks and tool iterations, because number of messages alone is not an adequate unit of resource consumption.

Question 200

A retrieval pipeline returns many neighboring pages that match keywords but not the intended question. Which implementation mechanism is needed?

  1. Retrieve relevant evidence per user question
  2. Use a task-specific worker for verbose analysis
  3. Filter irrelevant document chunks
  4. Put stable instructions ahead of volatile user material

Explanation

Correct Answer

C:

Relevance reranking can reduce noise while keeping evidence that supports the answer, subject to this requirement: excessive filtering can discard essential exceptions or contradictory passages.

Incorrect Answers

A:

Retrieval selects bounded passages pertinent to the current question and supplies their source context; this cannot substitute for the behavior that relevance reranking can reduce noise while keeping evidence that supports the answer, because selecting on superficial keywords alone may miss nuanced exceptions.

B:

Delegating bulky examination to an isolated worker can keep main context focused; in contrast, relevance reranking can reduce noise while keeping evidence that supports the answer, while the worker's compressed report must disclose uncertainties and source references.

D:

A stable prefix supports prompt-cache reuse and clearer separation of policy from per-request data; here, the controlling capability is that relevance reranking can reduce noise while keeping evidence that supports the answer, but if user text appears before the checkpoint it can invalidate shared cache hits.

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