Question 590
A research worker should search current public sources without possessing a production deployment credential. What would best address the stated failure mode?
- Use structured tool schemas for business actions
- Add a human gate for high-stakes changes
- Use a custom client tool for a private business operation
- Limit external tools to necessary operations
Explanation
Correct Answer
D:Minimal capabilities reduce the harm of prompt injection or mistaken planning; more available tools do not guarantee better task performance.
Incorrect Answers
A:Typed schemas and dispatcher checks convert an intention into a constrained operation; here, the controlling capability is that minimal capabilities reduce the harm of prompt injection or mistaken planning, but general Skill text cannot replace a server-enforced action interface.
B:Approval controls and audit records separate planning from privileged commitment; the missing capability is more specific: minimal capabilities reduce the harm of prompt injection or mistaken planning; a system prompt telling the model to act cautiously does not itself block execution.
C:A user-defined tool schema plus local dispatcher safely bridges the model to that system; by comparison, the intended solution relies on this: minimal capabilities reduce the harm of prompt injection or mistaken planning; the developer must operate the function and enforce its permissions.
Question 591
A low-risk conversational helper feels unresponsive unless users see text promptly. What is the correct engineering decision?
- Optimize for budget under a fixed accuracy floor
- Optimize for correctness before model speed
- Capture explicit nonfunctional latency targets
- Optimize for fast first visible output
Explanation
Correct Answer
D:Streaming and a suitable model can improve perceived responsiveness without changing the task's core accuracy bar; the implementation must still recognize that the first byte time and total completion time are different performance metrics.
Incorrect Answers
A:Candidate models are compared on both quality and total per-item spending across realistic inputs; this cannot substitute for the behavior that streaming and a suitable model can improve perceived responsiveness without changing the task's core accuracy bar, because selecting the cheapest model regardless of failures can increase downstream operational costs.
B:The acceptance process emphasizes measured quality and human review before pursuing latency reductions; in contrast, streaming and a suitable model can improve perceived responsiveness without changing the task's core accuracy bar, while a fast answer is not suitable if it violates the required error threshold.
C:A latency requirement defines measurable service-level timing and informs streaming and model selection; this addresses a separate concern, not the fact that streaming and a suitable model can improve perceived responsiveness without changing the task's core accuracy bar; correct semantic output does not prove the product meets a response-time constraint.
Question 592
A proprietary order system exposes a narrow internal lookup known only to the application's backend. Which solution should the engineer adopt?
- Use an MCP server for a reusable integration
- Use structured tool schemas for business actions
- Use a built-in server tool for a supported commodity task
- Use a custom client tool for a private business operation
Explanation
Correct Answer
D:A user-defined tool schema plus local dispatcher safely bridges the model to that system; the developer must operate the function and enforce its permissions.
Incorrect Answers
A:A protocol server exposes shared tools and resources with consistent discovery semantics; here, the controlling capability is that a user-defined tool schema plus local dispatcher safely bridges the model to that system, but it adds infrastructure and identity concerns compared with a single local function.
B:Typed schemas and dispatcher checks convert an intention into a constrained operation; the missing capability is more specific: a user-defined tool schema plus local dispatcher safely bridges the model to that system; general Skill text cannot replace a server-enforced action interface.
C:A built-in server tool uses the platform's execution flow and supported result types; by comparison, the intended solution relies on this: a user-defined tool schema plus local dispatcher safely bridges the model to that system; it does not grant custom private database access.
Question 593
Before running a risky shell command, the organization must execute an automatic validation outside the model's discretion. What change is best aligned with the goal?
- Use non-interactive print mode for CI automation
- Use a specialized subagent for focused review
- Use hooks for deterministic lifecycle checks
- Use a reusable Skill for repeatable know-how
Explanation
Correct Answer
C:Configured hooks invoke developer-controlled checks at defined event boundaries, subject to this requirement: a conversational instruction to perform the check is not equivalent to a guaranteed hook.
Incorrect Answers
A:Headless command-line invocation can run a prompt and return consumable output under controlled settings; this cannot substitute for the behavior that configured hooks invoke developer-controlled checks at defined event boundaries, because CI still needs noninteractive credentials and appropriately limited permissions.
B:A subagent delegates bounded work with appropriate tools and returns results to the parent; in contrast, configured hooks invoke developer-controlled checks at defined event boundaries, while the delegated agent should not inherit broader privileges than necessary.
D:A Skill packages reusable task guidance so the coding agent can apply a consistent procedure; here, the controlling capability is that configured hooks invoke developer-controlled checks at defined event boundaries, but guidance does not grant access to a tool or enforce backend authorization.
Question 594
An engineer is tempted to paste a production API token into a shared project configuration. What is the best fit for the described situation?
- Fail closed on missing safety configuration
- Keep secrets out of settings files committed to Git
- Document required environment variables
- Use organization-managed policy for enforced restrictions
Explanation
Correct Answer
B:Secrets belong in protected secret stores or environment injection rather than tracked source; notably, repository history can preserve credentials even after a later edit deletes them.
Incorrect Answers
A:Startup validation blocks unsafe execution rather than falling back to all permissions; the requirement instead turns on this behavior: secrets belong in protected secret stores or environment injection rather than tracked source; the absence of a guardrail must not silently widen access.
C:An explicit configuration contract makes setup repeatable without embedding secret values; in contrast, secrets belong in protected secret stores or environment injection rather than tracked source, while copying the original developer's entire environment is neither secure nor maintainable.
D:Managed settings supply centrally governed guardrails rather than relying on voluntary user configuration; this addresses a separate concern, not the fact that secrets belong in protected secret stores or environment injection rather than tracked source; ordinary project instructions are not equivalent to an enforced corporate security policy.
Question 595
A workflow resends the same large context after a gap longer than a short cache lifetime. Which design would meet this acceptance criterion?
- Inspect cache read and creation usage
- Place reusable material before per-user content
- Choose a supported model identifier
- Choose cache TTL based on reuse interval
Explanation
Correct Answer
D:A supported longer-duration cache may suit reuse that occurs after the default window; a crucial qualification is that longer cache writes can cost more, so economics depend on actual hit frequency.
Incorrect Answers
A:Cache usage metrics distinguish newly written content from tokens read from cache; in contrast, a supported longer-duration cache may suit reuse that occurs after the default window, while an assumed hit without usage evidence can hide prompt-prefix changes.
B:Organizing requests into a shared prefix and variable suffix improves the chance of cache hits; this addresses a separate concern, not the fact that a supported longer-duration cache may suit reuse that occurs after the default window; if the variable field appears before a checkpoint, the hashed prefix changes per request.
C:The model field identifies the deployed model; production should verify current IDs and compatibility; here, the controlling capability is that a supported longer-duration cache may suit reuse that occurs after the default window, but guessing an informal nickname can yield an invalid-model error or unexpected alias changes.
Question 596
A migration design requires weighing conflicting requirements and multiple constraints. Which choice should be used here?
- Benchmark thinking settings with task-specific tests
- Use extra reasoning on high-complexity tasks
- Verify model thinking compatibility
- Recognize latency due to extra reasoning
Explanation
Correct Answer
B:Supported thinking or effort settings allocate greater reasoning capacity to difficult deliberation; notably, added reasoning can increase latency and cost and still requires factual validation.
Incorrect Answers
A:Controlled quality and latency measurements show whether the chosen task benefits; the requirement instead turns on this behavior: supported thinking or effort settings allocate greater reasoning capacity to difficult deliberation; assuming maximal thinking is universally optimal wastes resources.
C:The integrator checks supported thinking parameters for the chosen version and request mode; in contrast, supported thinking or effort settings allocate greater reasoning capacity to difficult deliberation, while feature names and modes can differ between model releases.
D:Reasoning controls and task complexity affect total latency independently of answer length; this addresses a separate concern, not the fact that supported thinking or effort settings allocate greater reasoning capacity to difficult deliberation; reducing displayed text alone may not remove the underlying thinking cost.
Question 597
A policy assistant must cite a regulation revised after the model's training material. Which development approach should be selected?
- Handle context overflow before submission
- Use retrieval for current factual evidence
- Use deterministic validation after generation
- Treat output as generated prediction, not a database read
Explanation
Correct Answer
B:A retrieval step supplies updated source text before answer generation, subject to this requirement: retrieved content still needs authenticity and relevance checks.
Incorrect Answers
A:Chunking, retrieval, or compaction reduces input size while retaining decision-relevant evidence; in contrast, a retrieval step supplies updated source text before answer generation, while increasing output max_tokens cannot make an overlong input fit.
C:Business code checks numeric bounds independently of the model's prose; here, the controlling capability is that a retrieval step supplies updated source text before answer generation, but fluent explanations cannot override a failed validation rule.
D:The language model generates plausible text rather than querying a company's live inventory by itself; the missing capability is more specific: a retrieval step supplies updated source text before answer generation; it needs authorized retrieval to ground fast-changing private facts.
Question 598
A coding assistant reviews backend secrets and a public documentation task in parallel without cross-contamination. Which approach targets the deciding condition?
- Keep a structured conversation summary
- Remove stale tool outputs
- Use targeted tool requests instead of bulk dumps
- Isolate unrelated subtasks in separate contexts
Explanation
Correct Answer
D:Distinct subagent contexts prevent irrelevant data from filling the coordinator's main conversation, subject to this requirement: true confidentiality still depends on tool access and data-sharing controls.
Incorrect Answers
A:A summary stores salient decisions, constraints, and references in a reduced representation; the requirement instead turns on this behavior: distinct subagent contexts prevent irrelevant data from filling the coordinator's main conversation; compression may omit nuance and needs checking for critical obligations.
B:Pruning outdated results frees context for more relevant evidence and current questions; this cannot substitute for the behavior that distinct subagent contexts prevent irrelevant data from filling the coordinator's main conversation, because removing the sole authoritative citation would weaken the final answer.
C:Narrow queries reduce context consumption and sensitive data exposure simultaneously; in contrast, distinct subagent contexts prevent irrelevant data from filling the coordinator's main conversation, while returned results still need authorization before the query executes.
Question 599
A short-lived analysis needs intermediate notes but they must be discarded after the task ends. Which measure directly addresses the problem?
- Retry a failed read with bounded backoff
- Use ephemeral scratch context for one run
- Use a compact task summary for handoff
- Use keyed long-term memory for approved facts
Explanation
Correct Answer
B:Transient context supports local reasoning without creating long-term customer state; in production, it should not be mistaken for durable memory across new sessions.
Incorrect Answers
A:Retrying after increasing delays can recover a transient read failure without changing external state; the missing capability is more specific: transient context supports local reasoning without creating long-term customer state; the caller must stop after a limited number of attempts and surface persistent failure.
C:A narrow summary communicates findings, evidence, and unresolved questions within a smaller token budget; the needed capability is different—transient context supports local reasoning without creating long-term customer state—and it must not silently claim unverified findings are authoritative.
D:Stored records indexed by authorized account ID can be retrieved for later tasks, whereas the requirement calls for another behavior: transient context supports local reasoning without creating long-term customer state; access controls and deletion policy must govern persistence.
Question 600
A change to prompts and tool permissions could weaken production security if merged without scrutiny. Which solution should the engineer adopt?
- Require pull-request review
- Write a design review before implementation
- Maintain a change log with impact scope
- Write a rollback plan
Explanation
Correct Answer
A:Peer review examines diff, tests, and ownership before a controlled merge, although direct unreviewed edits reduce accountability and increase regression risk.
Incorrect Answers
B:Architecture review assesses interfaces, failure modes, and control ownership before code changes; however, the critical mechanism here is that peer review examines diff, tests, and ownership before a controlled merge, and implementing immediately can lock in unsafe assumptions and costly redesign.
C:Release records state compatibility, affected endpoints, and migration actions; the requirement instead turns on this behavior: peer review examines diff, tests, and ownership before a controlled merge; a model-generated summary alone does not define the actual deployed contract.
D:A rollback procedure identifies safe prior artifacts, migration constraints, and trigger thresholds; this cannot substitute for the behavior that peer review examines diff, tests, and ownership before a controlled merge, because reverting an application binary may be insufficient after incompatible schema changes.