Question 320
An MCP tool succeeds from one laptop but fails in CI because an expected environment variable is absent. What should the engineer check?
- Harden the transport with TLS and authenticate clients before sensitive tool calls
- Test tool exposure, permission filters and server-side authorization
- Align server identifier and tool-name changes with client configuration
- Verify required environment variables are injected in the CI runtime
Explanation
Correct Answer
D:An MCP configuration that expands environment references depends on that process's environment. Local developer shell settings do not automatically exist in pipeline runners.
Incorrect Answers
A:Tool inputs and results may contain confidential operational data; supporting protecting an MCP network connection cannot replace the required supplying MCP configuration variables in CI.
B:Connectivity alone does not grant every capability; this contributes to diagnosing a connected but unavailable MCP tool without meeting the need for supplying MCP configuration variables in CI.
C:Clients refer to discovered capabilities using identifiers; that serves migrating MCP tool identifiers safely, whereas the case requires supplying MCP configuration variables in CI.
Question 321
A developer made one tool with twenty optional arguments for every order workflow. What redesign improves selection?
- Recognize a stable idempotency key for mutation requests
- Expose a stable output schema and version incompatible changes
- Adopt an unambiguous timestamp representation with timezone requirements
- Split the overloaded interface into cohesive operations with narrow schemas
Explanation
Correct Answer
D:Smaller purpose-built tools make argument requirements and authorization easier to reason about. A giant optional parameter surface encourages invalid combinations and ambiguous dispatch.
Incorrect Answers
A:A client-provided operation identifier lets the service recognize repeat submissions and return the prior outcome; that may serve protecting a mutating tool against replay, but it is no substitute for reducing an overloaded tool schema.
B:A predictable contract lets agents and validators consume results consistently; this may support managing output schema compatibility but cannot alone meet the requirement for reducing an overloaded tool schema.
C:Machine-oriented time fields and explicit timezone semantics prevent interpretation drift; the stated problem requires reducing an overloaded tool schema rather than normalizing temporal tool parameters.
Question 322
A tool receives an HTTP 200 response that embeds a business-rule rejection. Which observability step makes the failed action visible?
- Formalize ordered responsibilities and nonconflicting hook contracts
- Mandate protected-path checks in a pre-tool permission handler
- Adopt a PreToolUse hook that denies matching unsafe invocations
- Inspect result payload semantics in a post-tool validation hook
Explanation
Correct Answer
D:Transport success does not necessarily mean business success. A validation hook can classify an application error in the response and report it to orchestration.
Incorrect Answers
A:Clear handler scope and explicit failure precedence prevent accidental interactions; this addresses resolving conflicting lifecycle handlers, while the question is about detecting business failure despite transport success.
B:A programmatic check can compare the requested path to restricted locations before access occurs; that relates to denying sensitive file reads before access, not the necessary detecting business failure despite transport success.
C:Although preToolUse runs before a tool call and can participate in the decision to block it, the choice addresses blocking disallowed commands before tool execution and not the requested detecting business failure despite transport success.
Question 323
An agent ignores a recently edited project convention because the working session has not loaded it. What should the operator verify?
- Write down the repository's actual build, lint and test commands with prerequisites
- Clarify the conflicting scope and establish one authoritative test command
- Confirm the intended CLAUDE.md is discovered and loaded for that working context
- Employ appropriately scoped instructions in the package's directory
Explanation
Correct Answer
C:Rules are useful only when the assistant encounters the relevant file and scope. Checking active instruction discovery is more direct than repeating the rule in unrelated prompts.
Incorrect Answers
A:Executable command knowledge is stable operational context; the unsatisfied requirement is verifying instruction-file loading, not recording project-specific validation commands.
B:Overlapping instructions can create inconsistent tool behavior; in this scenario, eliminating conflicting project instructions is secondary to verifying instruction-file loading.
D:When addressing selecting package-specific guidance, directory-local guidance gives specialized expectations without imposing them on unrelated packages; the present task instead calls for verifying instruction-file loading.
Question 324
A support ticket requires independent checks of entitlement and shipping state that share no mutation. Which orchestration improves latency?
- Run independent read-only specialist tasks concurrently and join their outputs
- Filter redundant outputs from planned search tasks and share a normalized evidence index
- Prioritize making synthesis depend on the completed evidence-review stage
- Forward each worker only the minimum task-specific context and authorized tools
Explanation
Correct Answer
A:When branches have no data dependency, parallel execution reduces waiting time while leaving the coordinator responsible for combining results and handling any branch failure.
Incorrect Answers
B:Centralizing a work plan and results catalogue prevents workers from independently repeating identical queries; avoiding redundant delegated searches and parallelizing independent evidence gathering are distinct responsibilities under these constraints.
C:A dependency edge enforces that the reporting worker receives validated source material; the action supports enforcing a dependency before synthesis but falls short of parallelizing independent evidence gathering in this case.
D:Specialization should come with least-necessary information and access; this can serve compartmentalizing delegated context and privileges, but leaves the requirement for parallelizing independent evidence gathering unmet.
Question 325
A worker shares the entire user transcript with a specialist that only needs one schema definition. What should change?
- Maintain a concise referenced policy finding and its source version
- Pass the minimum relevant schema and task constraints to that specialist
- Condense earlier turns into a verified summary that keeps decisions and open actions
- Carry forward pending approvals, blockers and next required actions explicitly
Explanation
Correct Answer
B:Context selection reduces noise and accidental disclosure. A narrowly scoped subtask should not inherit unrelated user history unless required for correctness.
Incorrect Answers
A:Cached validated findings reduce repetitive reads while retaining provenance; that serves avoiding repeated retrieval of unchanged evidence, whereas the case requires minimizing context passed to specialist workers.
C:Context compaction reduces token load while retaining essential state; its benefits for retaining critical state while compacting history do not resolve the need for minimizing context passed to specialist workers.
D:Pending workflow state is essential for safe resumption; supporting keeping unresolved control gates in condensed context cannot replace the required minimizing context passed to specialist workers.
Question 326
A rule is present but not triggered for nested paths because its pattern matches only the top directory. What should be corrected?
- Update the glob to cover nested directories explicitly
- Exercise rule glob matches against real repository paths
- Locate globally applicable policy in an always-loaded or enforced layer
- Divide language-specific guidance into separate path-targeted rule files
Explanation
Correct Answer
A:Path patterns control rule discovery. A pattern that misses deeper modules leaves their edits without the intended local guidance.
Incorrect Answers
B:A simple match audit reveals dead rules that never load; a method suited to verifying rule selectors actually match files does not meet the need for matching nested paths in a rule glob.
C:For keeping organization-wide policy outside conditional globs, a conditional file rule may not activate in every session; however, this requirement concerns matching nested paths in a rule glob.
D:Distinct rules with distinct globs avoid contradictory language conventions; that may serve separating language-specific rules, but it is no substitute for matching nested paths in a rule glob.
Question 327
A batch may remain unfinished when its processing window expires. What should the system plan for?
- Store results to controlled storage within the documented retrieval period
- Detect expiration and resubmit eligible unfinished requests separately
- Observe pending request limits and pace submissions by available capacity
- Delete streaming from batch requests and process completed result artifacts
Explanation
Correct Answer
B:Deferred processing has time limits and can return expired outcomes. Recovery must identify unfinished work instead of assuming eventual success indefinitely.
Incorrect Answers
A:Although batch result URLs are not an archival store, the choice addresses preserving batch output beyond service retention and not the requested recovering expired batch requests.
C:Batch queues remain subject to provider limits and account constraints; the option would serve respecting throughput limits when batching, not recovering expired batch requests as specified in the stem.
D:Batch execution yields collected request outcomes rather than an interactive token stream; this addresses using supported response delivery for batch processing, while the question is about recovering expired batch requests.
Question 328
A developer sends only the text of a prior assistant tool request with a user tool_result on the next API call. What context is missing?
- Continue with a prose-only summary of the previous assistant message while omitting its structured tool_use block
- Store the tool implementation in the authorized execution environment
- Preserve the assistant tool_use content block as part of the message history
- Direct the refusal through the approved response and escalation policy
Explanation
Correct Answer
C:The next request needs the structured assistant invocation and the matching tool result in a coherent transcript. A plain-text paraphrase cannot reliably supply the expected call linkage.
Incorrect Answers
A:Tool results must be matched to the actual tool_use call identifiers in the assistant message history. A text summary drops the structural linkage required by the tool protocol.
B:When addressing distinguishing model intent from tool execution, client tools are interfaces advertised to the model; the hosting application runs their operations; the present task instead calls for preserving the structured conversation transcript.
D:A refusal is a distinct termination state; the unsatisfied requirement is preserving the structured conversation transcript, not handling refusal as a distinct state.
Question 329
An agent repeatedly calls a costly external tool even after reaching a configured quota. Which enforcement point is useful?
- Reject over-budget requests in the tool-call authorization boundary
- Record the tool completion event and record its returned outcome
- Begin by attaching the validation to the agent-turn or stop lifecycle event
- Block closed or suspend the protected action according to explicit policy
Explanation
Correct Answer
A:Tracking usage and checking it before execution makes a budget enforceable. An after-the-fact warning would not prevent the charge-producing invocation.
Incorrect Answers
B:Post-execution telemetry can contain the result and status of a tool operation; this can serve auditing tool execution outcomes, but leaves the requirement for enforcing a quota at the tool boundary unmet.
C:End-of-turn hooks can trigger aggregate checks that would be wasteful after every individual edit; that addresses validating at the end of agent activity, whereas this case calls for enforcing a quota at the tool boundary.
D:When the decision service is unavailable, a high-risk action should not silently proceed; this concerns handling a failed mandatory policy check, leaving the requirement for enforcing a quota at the tool boundary outstanding.
Question 330
A company wants only one repository to use a database migration skill, not every personal project. Which placement is appropriate?
- Store the release workflow as a project skill with clear instructions
- Retain supporting resources within the skill and load them when needed
- Store the skill at project scope rather than globally for the user
- Constrain skill tool permissions to the necessary read-oriented operations
Explanation
Correct Answer
C:Scope limits where reusable workflows are discovered and available. A migration operation is repository-specific and should not automatically appear in unrelated work.
Incorrect Answers
A:A reusable skill can collect steps, supporting files and tool permissions under one invocable workflow; its benefits for making a repeatable release workflow reusable do not resolve the need for limiting a sensitive skill to its repository.
B:Skills can bundle reference material and executable helpers for a focused task; supporting packaging on-demand skill resources cannot replace the required limiting a sensitive skill to its repository.
D:A skill's allowed tool surface should match its declared job; that serves defining least-privilege tools for a skill, whereas the case requires limiting a sensitive skill to its repository.