Question 240
A long-lived support session stores all retrieved customer records forever. Which retention practice reduces risk?
- Retain a session identifier and resume its authorized conversation state
- Cross-check the fork against the current codebase and rerun relevant checks
- Persist only necessary state and apply an explicit retention or deletion policy
- Verify the user's session privileges again before protected work continues
Explanation
Correct Answer
C:Resumability does not justify indefinite storage of sensitive context. State minimization and lifecycle controls balance continuity with privacy obligations.
Incorrect Answers
A:Because session continuity reconnects the agent to relevant history and progress, the measure supports resuming a saved agent session rather than the needed limiting persisted session data.
B:A successful result on an older branch does not prove current compatibility; its benefits for validating branch changes before merge do not resolve the need for limiting persisted session data.
D:A stored conversation is not a perpetual permission grant; this contributes to revalidating authorization on session resume without meeting the need for limiting persisted session data.
Question 241
A code reviewer returns subjective complaints as production blockers. How should the prompt distinguish severity?
- Define severity levels by reproducible impact and required evidence
- Give a clear output structure with explicit labels for evidence and risk rationale
- Declare the defect categories and require a cited code location for every finding
- Ascribe decisions and proposals to named sources, preserving unresolved disagreement
Explanation
Correct Answer
A:Rubric-based severity promotes consistent triage. Cosmetic preferences should not be treated as equivalent to security, correctness or data-loss defects.
Incorrect Answers
B:For standardizing the structure of risk assessments, combining an evaluation rubric with an output contract makes responses comparable; however, this requirement concerns scoring review findings against a severity rubric.
C:Evidence-based criteria distinguish observed defects from speculative criticism; that may serve preventing unsupported PR review findings, but it is no substitute for scoring review findings against a severity rubric.
D:Attribution is part of factual correctness; this may support preserving attribution in generated meeting summaries but cannot alone meet the requirement for scoring review findings against a severity rubric.
Question 242
Claude Code has mapped a feature, but tests reveal unexpected shared state. What should it do before continuing to edit broadly?
- Detail deployment order, checkpoints and rollback triggers before execution
- Review references and present a removal plan before deletion
- Focus on returning to investigation and revise the plan around the newly discovered dependency
- Restrict the workflow to planning rather than applying edits
Explanation
Correct Answer
C:Planning is iterative when evidence changes. Continuing the initial implementation blindly ignores a material assumption that no longer holds.
Incorrect Answers
A:Multi-service updates have dependencies and failure domains that cannot be improvised safely; the deciding condition is replanning when new dependency evidence appears, whereas this action concerns preparing release sequencing and rollback.
B:Deleting files on a hunch can break hidden consumers; the option would serve researching destructive refactors before execution, not replanning when new dependency evidence appears as specified in the stem.
D:An assessment boundary prohibits side effects regardless of how confident the proposed fix appears; that relates to honoring a read-only operating constraint, not the necessary replanning when new dependency evidence appears.
Question 243
An agent is told to inspect all test files matching a filename convention but does not know their paths. Which built-in operation helps first?
- Adopt a scoped Edit with an exact matching target
- Launch the formatter through Bash and inspect the resulting diff or file changes
- Use Glob to enumerate files matching the path pattern
- Locate with Grep occurrences, inspect affected files, then apply controlled Edits
Explanation
Correct Answer
C:Glob resolves matching file paths by name or path shape. It avoids opening every directory or scanning file contents before the candidate files are known.
Incorrect Answers
A:When addressing making a minimal configuration edit, a targeted replacement can preserve neighboring configuration content; the present task instead calls for enumerating files by filename pattern.
B:Process output alone may not prove exactly what changed; verifying changes produced by an executable command and enumerating files by filename pattern require different controls in this situation.
D:A search-first workflow reveals scope and avoids blind global replacement; in this scenario, scoping a multi-file refactor before mutation is secondary to enumerating files by filename pattern.
Question 244
A team repeats formatting rules in every prompt while a common repository convention already exists. What improves consistency?
- Refresh version-sensitive project instructions alongside the framework migration
- Encode the stable rules once in reviewed project instructions
- Choose appropriately scoped instructions in the package's directory
- Maintain mandatory security policy in enforced controls rather than editable personal preferences
Explanation
Correct Answer
B:Persistent project guidance makes the same expectations available across sessions. Repeating them manually creates drift and consumes prompt space.
Incorrect Answers
A:Standing guidance can become stale as dependencies and conventions evolve; this can serve maintaining instruction accuracy after upgrades, but leaves the requirement for persisting stable codebase conventions unmet.
C:Directory-local guidance gives specialized expectations without imposing them on unrelated packages; this concerns selecting package-specific guidance, leaving the requirement for persisting stable codebase conventions outstanding.
D:Instruction hierarchy should not be the sole defense for high-risk operations; separating organization policy from user preferences and persisting stable codebase conventions are distinct responsibilities under these constraints.
Question 245
A form parser reports duplicate items because two pages overlap in the scan. What should post-processing do?
- Deduplicate extracted records using source identifiers and document context
- Enforce domain-specific calculation and policy rules after schema parsing
- Enforce a bounded retry budget and move the document to an exception queue
- Keep the original string and flag ambiguity until locale evidence resolves it
- Resubmit with targeted feedback identifying the invalid field and expected format
Explanation
Correct Answer
A:Overlapping page captures can lead to double counting. Traceable source positions and stable keys permit safe duplicate detection before posting totals.
Incorrect Answers
B:Schema guarantees shape, not compliance with business calculations; this contributes to validating business invariants after schema checks without meeting the need for removing duplicate extraction artifacts.
C:Repeated retries cost money and may not improve fidelity; that serves bounding extraction repair attempts, whereas the case requires removing duplicate extraction artifacts.
D:Because a date parser cannot safely invent locale rules, the measure supports handling ambiguous date interpretations rather than the needed removing duplicate extraction artifacts.
E:Feedback-based repair gives the model a specific correctable defect; its benefits for targeting a retry at a correctable parse defect do not resolve the need for removing duplicate extraction artifacts.
Question 246
Two rules apply to the same file and disagree about whether tests are mandatory. What should maintainers do?
- Formalize path-scoped rules under .claude/rules with file-pattern applicability
- Resolve overlapping conditions and specify one unambiguous policy
- Position globally applicable policy in an always-loaded or enforced layer
- Change the glob to cover nested directories explicitly
Explanation
Correct Answer
B:Conflicting active instructions can make outcomes nondeterministic. Reducing overlap or reconciling requirements is better than hoping the model guesses priority correctly.
Incorrect Answers
A:Path-conditioned repository guidance applies specialized instructions only to matching files; that may serve applying code-style rules by path, but it is no substitute for preventing contradictory conditional instructions.
C:A conditional file rule may not activate in every session; the stated problem requires preventing contradictory conditional instructions rather than keeping organization-wide policy outside conditional globs.
D:Path patterns control rule discovery; a method suited to matching nested paths in a rule glob does not meet the need for preventing contradictory conditional instructions.
Question 247
A team ships a skill with an unclear name and no description, so developers rarely use it. What metadata improvement helps?
- Exclude embedded secrets and access approved credentials through secure runtime configuration
- Document the skill as plan analysis and deny deployment or mutation tools
- Select project-relative paths and document required environment inputs
- Give the skill a clear trigger description and intended use cases
Explanation
Correct Answer
D:Descriptive metadata improves discovery and correct invocation. A generic label leaves humans and agents unsure which workflow is appropriate.
Incorrect Answers
A:Skill contents may be loaded into model context and shared in repositories; this addresses securing reusable skill contents, while the question is about improving skill discoverability.
B:Skill language and access control should agree; that relates to separating review skills from deployment skills, not the necessary improving skill discoverability.
C:Although portable skills avoid machine-specific assumptions and resolve resources from a predictable project context, the choice addresses removing machine-specific assumptions from a skill and not the requested improving skill discoverability.
Question 248
A remote MCP service fails, but the coordinator cannot tell which tool invocation caused it. What field is useful?
- Emit a non-retryable validation error with the offending field identified
- Attach a trace or correlation identifier to the diagnostic result
- Label the outcome as uncertain and request reconciliation before replay
- Replay time-sensitive reads after recovery and replace stale failure state
Explanation
Correct Answer
B:Correlation metadata links user-facing failure to server logs and specific attempts. This makes distributed troubleshooting possible without including sensitive log payloads.
Incorrect Answers
A:Input-shape errors persist until arguments change; the unsatisfied requirement is correlating errors across distributed tool services, not distinguishing validation faults from transient errors.
C:When considering handling an ambiguous post-request timeout, a timeout does not prove nonexecution for mutating operations; the requirement for correlating errors across distributed tool services remains unmet.
D:When addressing refreshing state after a transient service failure, cached failure artifacts are not live evidence; the present task instead calls for correlating errors across distributed tool services.
Question 249
A production team wants one remotely hosted MCP service reachable by several controlled clients. Which connection pattern fits?
- Configure a remote HTTP-based MCP endpoint with appropriate authentication
- Assess and approve the project server before granting it access
- Audit tool exposure, permission filters and server-side authorization
- Validate required environment variables are injected in the CI runtime
Explanation
Correct Answer
A:Networked MCP transport supports a centrally hosted integration across clients. A locally launched child process is not the same deployment model.
Incorrect Answers
B:Repository-supplied configurations can execute processes or contact services; reviewing untrusted project MCP definitions and connecting to a shared remote MCP service are distinct responsibilities under these constraints.
C:Connectivity alone does not grant every capability; the action supports diagnosing a connected but unavailable MCP tool but falls short of connecting to a shared remote MCP service in this case.
D:An MCP configuration that expands environment references depends on that process's environment; this can serve supplying MCP configuration variables in CI, but leaves the requirement for connecting to a shared remote MCP service unmet.
Question 250
A recursive support agent repeatedly asks for new tools without approaching a resolution. Which host control prevents unbounded execution?
- Advance the paused assistant turn with its response content preserved
- Send the refusal through the approved response and escalation policy
- Safeguard the tool implementation in the authorized execution environment
- Enforce a step budget and a defined fallback when the loop exceeds it
Explanation
Correct Answer
D:Independent limits on turns, time, and side effects prevent runaway agent cycles. Exceeding a budget should trigger a controlled handoff or incomplete status, not infinite recursion.
Incorrect Answers
A:A pause_turn can mark a server-tool iteration boundary; that serves continuing a paused server-tool loop, whereas the case requires bounding an autonomous agent loop.
B:Because a refusal is a distinct termination state, the measure supports handling refusal as a distinct state rather than the needed bounding an autonomous agent loop.
C:Client tools are interfaces advertised to the model; the hosting application runs their operations; its benefits for distinguishing model intent from tool execution do not resolve the need for bounding an autonomous agent loop.