Question 460
A diagnosis requires an agent to inspect error logs and adapt its following query to what it finds. Which engineering practice is the strongest match?
- Ask a clarifying question when critical input is missing
- Observe a tool result before choosing the next action
- Detect and resolve conflicting worker findings
- Escalate after repeated inconclusive tool results
Explanation
Correct Answer
B:Alternating action and observation permits evidence-driven choices rather than a fully predetermined plan, subject to this requirement: every loop must check for stop conditions and tool failures.
Incorrect Answers
A:Requesting an authorized scope prevents guesses from becoming unintended side effects; in contrast, alternating action and observation permits evidence-driven choices rather than a fully predetermined plan, while an internal plan cannot legitimately invent the missing tenancy information.
C:The coordinator compares sources, dates, and evidence before emitting a final claim; here, the controlling capability is that alternating action and observation permits evidence-driven choices rather than a fully predetermined plan, but voting by worker count is not a substitute for checking authoritative documentation.
D:Bounded fallback returns uncertainty and a trace for human follow-up rather than fabricating certainty; the missing capability is more specific: alternating action and observation permits evidence-driven choices rather than a fully predetermined plan; extra iterations are not justified once the cost and confidence guardrails trigger.
Question 461
A research assistant has accumulated bulky transcripts but needs only verified decisions for its next stage. Which design would meet this acceptance criterion?
- Use a summarizing worker for context compression
- Use a specialist subagent
- Assign tool side effects to an execution worker
- Delegate retrieval to a research worker
Explanation
Correct Answer
A:A dedicated summarizer condenses retained facts with provenance and open issues for the parent context; in production, compression may omit nuance unless the system checks critical commitments and references.
Incorrect Answers
B:An isolated worker handles a bounded analysis and returns a concise finding to the parent; the needed capability is different—a dedicated summarizer condenses retained facts with provenance and open issues for the parent context—and the subagent still needs a scoped tool set and cannot silently inherit unlimited production privileges.
C:A tightly permissioned executor maps authorized operations into deterministic tool calls, whereas the requirement calls for another behavior: a dedicated summarizer condenses retained facts with provenance and open issues for the parent context; broad access in the planning component defeats the intended separation of duties.
D:A research worker gathers references and returns a compact evidence set to the lead; however, the critical mechanism here is that a dedicated summarizer condenses retained facts with provenance and open issues for the parent context, and it should not be allowed to publish the final legal conclusion as an independent authority.
Question 462
An agent can propose resource deletion but actual execution needs accountable human authorization. Which change would satisfy the constraint?
- Use a sandbox for model-requested code execution
- Add a human gate for high-stakes changes
- Limit external tools to necessary operations
- Combine a guarded custom tool with a human approval step
Explanation
Correct Answer
B:Approval controls and audit records separate planning from privileged commitment; its operational boundary is that a system prompt telling the model to act cautiously does not itself block execution.
Incorrect Answers
A:Process and network isolation contain potential side effects of generated programs; this cannot substitute for the behavior that approval controls and audit records separate planning from privileged commitment, because a model's internal safety judgment is not an OS permission boundary.
C:Minimal capabilities reduce the harm of prompt injection or mistaken planning; this addresses a separate concern, not the fact that approval controls and audit records separate planning from privileged commitment; more available tools do not guarantee better task performance.
D:The tool executes a constrained operation after an enforced business approval gate; here, the controlling capability is that approval controls and audit records separate planning from privileged commitment, but model-generated consent text is not an auditable approval event.
Question 463
A deployment changes request persistence and might need reversal under load. What is the appropriate next design choice?
- Keep rollback artifacts and data migration notes
- Use a compatibility test matrix for upgrades
- Use a sandbox backend for tool integration tests
- Document breaking API changes for callers
Explanation
Correct Answer
A:Release planning preserves prior binaries and any required data conversion strategy, although reverting only source files may not undo incompatible database changes.
Incorrect Answers
B:Matrix tests exercise supported combinations against contract and quality checks; however, the critical mechanism here is that release planning preserves prior binaries and any required data conversion strategy, and assuming backward compatibility without testing risks hidden behavioral changes.
C:A nonproduction environment exercises authorization, payload mapping, and failures safely; the requirement instead turns on this behavior: release planning preserves prior binaries and any required data conversion strategy; faking every backend call can hide actual integration incompatibilities.
D:A versioned change notice and migration path help clients adopt the new contract safely; this cannot substitute for the behavior that release planning preserves prior binaries and any required data conversion strategy, because a silent rollout may turn valid old requests into failures.
Question 464
A release needs one complete invoice-to-summary path through API, validation, and logging. Which solution matches the actual constraint?
- Benchmark baseline latency and cost
- Use a threat model before tool rollout
- Identify external integration contracts
- Implement a narrow vertical slice
Explanation
Correct Answer
D:A small end-to-end slice tests integration boundaries and production assumptions together, although building many disconnected components first delays discovery of interface mismatches.
Incorrect Answers
A:Capturing tokens, response times, failure rate, and quality before changes makes comparisons meaningful; the missing capability is more specific: a small end-to-end slice tests integration boundaries and production assumptions together; subjective demos cannot quantify a regression from the original system.
B:Mapping assets, adversaries, trust boundaries, and side effects exposes what must be protected; by comparison, the intended solution relies on this: a small end-to-end slice tests integration boundaries and production assumptions together; security should not be postponed until the tool can already send messages.
C:Interface requirements document endpoints, authentication, payloads, limits, and failure handling; the needed capability is different—a small end-to-end slice tests integration boundaries and production assumptions together—and a prose-level prompt is not a substitute for a validated service contract.
Question 465
A catalog search responds with 50 large product records but the model needs price and availability for three candidates. What should the team do at this decision point?
- Include stable identifiers in tool results
- Return only the fields necessary for a decision
- Preserve tool error codes separately from text
- Return a matching tool_result block
Explanation
Correct Answer
B:Structured filtering reduces context load and unnecessary data disclosure; notably, the model should not receive personal or unrelated fields merely because they exist upstream.
Incorrect Answers
A:IDs let downstream operations refer unambiguously to retrieved business objects; the requirement instead turns on this behavior: structured filtering reduces context load and unnecessary data disclosure; human-readable names alone may collide across tenants or records.
C:Typed errors guide safe retries, clarification, or escalation without inventing data; in contrast, structured filtering reduces context load and unnecessary data disclosure, while free-form error prose can be misread or stripped by downstream clients.
D:The following turn includes a result linked by tool_use_id so the model can continue coherently; this addresses a separate concern, not the fact that structured filtering reduces context load and unnecessary data disclosure; returning an unrelated human message loses the execution association.
Question 466
Engineers want a concise invocation for a repeated repository task such as a release-note draft. Which action most directly closes the gap?
- Use a specialized subagent for focused review
- Limit shell command authority in automation
- Use hooks for deterministic lifecycle checks
- Use a custom slash command for a named recurring action
Explanation
Correct Answer
D:A named command exposes reusable prompt-driven workflow entry points inside Claude Code, subject to this requirement: a command definition still depends on the session's tool permissions and context.
Incorrect Answers
A:A subagent delegates bounded work with appropriate tools and returns results to the parent; the requirement instead turns on this behavior: a named command exposes reusable prompt-driven workflow entry points inside Claude Code; the delegated agent should not inherit broader privileges than necessary.
B:Restricted command/tool permissions reduce the possible effects of prompt errors or malicious code comments; this cannot substitute for the behavior that a named command exposes reusable prompt-driven workflow entry points inside Claude Code, because using a temporary checkout is helpful but does not itself revoke external credentials.
C:Configured hooks invoke developer-controlled checks at defined event boundaries; in contrast, a named command exposes reusable prompt-driven workflow entry points inside Claude Code, while a conversational instruction to perform the check is not equivalent to a guaranteed hook.
Question 467
A brainstorming assistant should offer diverse headline candidates rather than repeatedly using similar phrasing. Which action most directly closes the gap?
- Use a more exploratory sampling setting when supported
- Use low-temperature settings when supported for consistency
- Verify model thinking compatibility
- Avoid assuming byte-identical responses
Explanation
Correct Answer
A:Sampling parameters can trade predictability for variety in compatible model modes—with this limitation: increased randomness may degrade factual precision and schema reliability.
Incorrect Answers
B:Lower randomness can reduce sampling variability in supported configurations; this cannot substitute for the behavior that sampling parameters can trade predictability for variety in compatible model modes, because it does not guarantee deterministic output or eliminate model errors.
C:The integrator checks supported thinking parameters for the chosen version and request mode; in contrast, sampling parameters can trade predictability for variety in compatible model modes, while feature names and modes can differ between model releases.
D:Generative sampling and platform execution can cause output variability, so compare validated properties instead; this addresses a separate concern, not the fact that sampling parameters can trade predictability for variety in compatible model modes; a hard textual hash is usually an unsuitable semantic acceptance metric.
Question 468
A project retains obsolete feature flags whose behavior is unclear after multiple releases. What should the team do at this decision point?
- Remove deprecated configuration under change control
- Use automated integration tests before release
- Review post-incident root causes
- Categorize and triage production incidents
Explanation
Correct Answer
A:A controlled cleanup traces dependencies, verifies current usage, and deletes unused flags with tests; its operational boundary is that removing all old switches at once can unintentionally disable required paths.
Incorrect Answers
B:Integration tests check how independently coded components actually interact under realistic conditions; this addresses a separate concern, not the fact that a controlled cleanup traces dependencies, verifies current usage, and deletes unused flags with tests; unit tests of isolated helpers cannot reveal all cross-component contract failures.
C:Root-cause analysis traces initiating conditions and systemic prevention beyond short-term restoration; here, the controlling capability is that a controlled cleanup traces dependencies, verifies current usage, and deletes unused flags with tests, but a successful restart does not explain why the same fault reappears.
D:Incident classification routes problems to the correct service, integration, or prompt team; the missing capability is more specific: a controlled cleanup traces dependencies, verifies current usage, and deletes unused flags with tests; treating every failure as bad model reasoning conceals infrastructure issues.
Question 469
A status classifier needs to return a specialized JSON mapping that is hard to infer from the label names. What is the correct engineering decision?
- Give a correct representative example
- Include a counterexample with correction
- Construct a representative validation set
- Test prompt examples for representativeness
Explanation
Correct Answer
A:A compact demonstration clarifies the intended input-to-output mapping; notably, examples should match the required schema and not use private customer records.
Incorrect Answers
B:Showing an unacceptable output and its corrected form highlights a specific formatting mistake; in contrast, a compact demonstration clarifies the intended input-to-output mapping, while a negative example without the correct alternative may create confusion.
C:A labeled suite of normal, edge, and adversarial examples measures practical performance; this addresses a separate concern, not the fact that a compact demonstration clarifies the intended input-to-output mapping; one anecdotal success does not establish production reliability.
D:Broader cases reveal where examples bias the model toward a narrow pattern; here, the controlling capability is that a compact demonstration clarifies the intended input-to-output mapping, but more examples do not automatically fix poor dataset coverage.
Question 470
A desktop application can call an MCP server locally while another runtime uses a remote server. Which approach targets the deciding condition?
- Use stdio for a local subprocess integration
- Use Streamable HTTP for remote MCP access
- Use a client-side transport adapter
- Expose a custom API adapter outside MCP when local-only
Explanation
Correct Answer
C:Transport abstraction lets the same protocol-level concepts operate over supported connection methods, although a transport switch does not erase server deployment or security requirements.
Incorrect Answers
A:Standard input and output suit tightly coupled local client-server interactions; by comparison, the intended solution relies on this: transport abstraction lets the same protocol-level concepts operate over supported connection methods; the server's standard output must remain protocol-clean instead of mixing human debug prints.
B:A network transport supports remote requests under explicit authentication and deployment controls; the needed capability is different—transport abstraction lets the same protocol-level concepts operate over supported connection methods—and opening an HTTP endpoint without access restrictions can expose powerful tools publicly.
D:A direct adapter may be simpler than running and maintaining a separate protocol server; however, the critical mechanism here is that transport abstraction lets the same protocol-level concepts operate over supported connection methods, and choosing MCP solely for naming consistency introduces unnecessary lifecycle work.