Question 10
For a nightly code quality job, acceptance depends on demonstrating that resuming a session does not bypass present-day authorization. Which approach is the strongest fit?
- treat old approval messages as unlimited authorization for later writes
- preserve both branch identifiers and outcome summaries
- resume the stored session by its identifier
- recheck permissions and current resource state before new side effects
- initialize a fresh session with a verified structured summary
Explanation
Correct Answer
D:For a nightly code quality job, conversation continuity is not a replacement for live access control or fresh tool state; this directly establishes that resuming a session does not bypass present-day authorization instead of relying on an unrelated workflow control.
Incorrect Answers
A:In a nightly code quality job, permissions and operational context can change between sessions; as a result, the stated requirement that resuming a session does not bypass present-day authorization would still not be met.
B:Although tracking lineage makes result comparison possible without confusing which session produced an artifact, that only supports the distinct goal that an exploratory branch can be compared with a baseline later, not the requirement that resuming a session does not bypass present-day authorization in a nightly code quality job.
C:Although session resumption retains the recorded conversation and avoids rebuilding the entire task from scratch, that only supports the distinct goal that a developer returns to an interrupted session with prior reasoning, not the requirement that resuming a session does not bypass present-day authorization in a nightly code quality job.
E:Although a bounded handoff transfers necessary facts without carrying discarded trial-and-error context, that only supports the distinct goal that a new agent gets only durable context from a completed job, not the requirement that resuming a session does not bypass present-day authorization in a nightly code quality job.
Question 11
Production readiness for an automated CI review service depends on ensuring that a common security rule remains active across the project and that reviewer feedback points to a specific missed requirement. Select the most appropriate combined approach.
- implement only turn the feedback into a concrete test or change request; make no corresponding change for the first requirement
- keep project-wide security guidance in shared instructions rather than a narrow path glob; and turn the feedback into a concrete test or change request
- scope the security instruction only to one unrelated test directory; and turn the feedback into a concrete test or change request
- scope the security instruction only to one unrelated test directory; and repeat the original prompt verbatim and expect a different outcome
- keep project-wide security guidance in shared instructions rather than a narrow path glob; and repeat the original prompt verbatim and expect a different outcome
Explanation
Correct Answer
B:In an automated CI review service, cross-cutting obligations should not disappear when an agent changes directories; alongside this, actionable feedback narrows the next edit and makes improvement measurable, satisfying the distinct requirements that a common security rule remains active across the project and that reviewer feedback points to a specific missed requirement.
Incorrect Answers
A:In an automated CI review service, actionable feedback narrows the next edit and makes improvement measurable, but the independent mechanism for the other criterion is absent: cross-cutting obligations should not disappear when an agent changes directories; as written, the design does not ensure that a common security rule remains active across the project.
C:In an automated CI review service, actionable feedback narrows the next edit and makes improvement measurable, but other code paths would lack the universal control; therefore the separate requirement that a common security rule remains active across the project would remain unsatisfied.
D:In an automated CI review service, other code paths would lack the universal control and unchanged inputs do not target the identified deficiency; the design would fail both conditions: a common security rule remains active across the project and reviewer feedback points to a specific missed requirement.
E:In an automated CI review service, cross-cutting obligations should not disappear when an agent changes directories, but unchanged inputs do not target the identified deficiency; therefore the separate requirement that reviewer feedback points to a specific missed requirement would remain unsatisfied.
Question 12
Before expanding a test failure triage bot, the owner must demonstrate that path-scoped rules are loaded for all intended source directories, while also demonstrating that the reviewer does not simply echo the generator's confidence. Which design provides both controls?
- add more prose telling the model to remember the missing rule; and give the reviewer the source and rubric, not only the draft's self-rating
- update the rule's paths glob to include the actual target paths; and give the reviewer the source and rubric, not only the draft's self-rating
- update the rule's paths glob to include the actual target paths; and provide the generator's score as the sole review criterion
- add more prose telling the model to remember the missing rule; and provide the generator's score as the sole review criterion
- implement only update the rule's paths glob to include the actual target paths; make no corresponding change for the second requirement
Explanation
Correct Answer
B:In a test failure triage bot, correct glob coverage determines whether conditional guidance is in context; alongside this, independent evidence lets a critic challenge unsupported assertions, satisfying the distinct requirements that path-scoped rules are loaded for all intended source directories and that the reviewer does not simply echo the generator's confidence.
Incorrect Answers
A:In a test failure triage bot, independent evidence lets a critic challenge unsupported assertions, but instructions outside the matched scope still will not be loaded; therefore the separate requirement that path-scoped rules are loaded for all intended source directories would remain unsatisfied.
C:In a test failure triage bot, correct glob coverage determines whether conditional guidance is in context, but anchoring on a self-assigned grade hides actual defects; therefore the separate requirement that the reviewer does not simply echo the generator's confidence would remain unsatisfied.
D:In a test failure triage bot, instructions outside the matched scope still will not be loaded and anchoring on a self-assigned grade hides actual defects; the design would fail both conditions: path-scoped rules are loaded for all intended source directories and the reviewer does not simply echo the generator's confidence.
E:In a test failure triage bot, correct glob coverage determines whether conditional guidance is in context, but the independent mechanism for the other criterion is absent: independent evidence lets a critic challenge unsupported assertions; as written, the design does not ensure that the reviewer does not simply echo the generator's confidence.
Question 13
Before expanding a customer refund workflow, the owner must demonstrate that client-side tool requests trigger another model turn after execution, while also demonstrating that each tool interface has a single clear operation boundary. Which design provides both controls?
- end the loop whenever assistant text appears; and separate read-only lookup from side-effecting update interfaces
- branch on stop_reason tool_use and execute returned tool_use blocks; and separate read-only lookup from side-effecting update interfaces
- end the loop whenever assistant text appears; and use a single universal tool that infers read versus write from prose
- branch on stop_reason tool_use and execute returned tool_use blocks; and use a single universal tool that infers read versus write from prose
Explanation
Correct Answer
B:In a customer refund workflow, the response explicitly signals pending client execution and the loop must provide results before another model decision; alongside this, clear operation boundaries support least privilege and understandable approvals, satisfying the distinct requirements that client-side tool requests trigger another model turn after execution and that each tool interface has a single clear operation boundary.
Incorrect Answers
A:In a customer refund workflow, clear operation boundaries support least privilege and understandable approvals, but text may accompany a tool request and is not the completion signal; therefore the separate requirement that client-side tool requests trigger another model turn after execution would remain unsatisfied.
C:In a customer refund workflow, text may accompany a tool request and is not the completion signal and overloading hides effects and makes execution review harder; the design would fail both conditions: client-side tool requests trigger another model turn after execution and each tool interface has a single clear operation boundary.
D:In a customer refund workflow, the response explicitly signals pending client execution and the loop must provide results before another model decision, but overloading hides effects and makes execution review harder; therefore the separate requirement that each tool interface has a single clear operation boundary would remain unsatisfied.
Question 14
The architect for a returns-policy assistant needs to ensure that persistent ambiguous data receives human review. What should the team implement?
- record validation failure codes and field paths per attempt
- keep retrying until the model finally emits something that passes
- stop after defined retries and route unresolved cases to a reviewer
- validate parsed data against application rules and source constraints
Explanation
Correct Answer
C:For a returns-policy assistant, a bounded escalation path prevents infinite loops and invented resolutions; this directly establishes that persistent ambiguous data receives human review instead of relying on an unrelated workflow control.
Incorrect Answers
A:Although granular observability reveals schema weaknesses and document patterns, that only supports the distinct goal that the team can diagnose which fields repeatedly fail extraction, not the requirement that persistent ambiguous data receives human review in a returns-policy assistant.
B:In a returns-policy assistant, unlimited retries waste resources and can reinforce guesses; as a result, the stated requirement that persistent ambiguous data receives human review would still not be met.
D:Although local validation detects missing cross-field relationships even when the JSON is syntactically valid, that only supports the distinct goal that an extraction error is caught before storage, not the requirement that persistent ambiguous data receives human review in a returns-policy assistant.
Question 15
An audit of a deployment-readiness verification process asks how to make sure that conflicting guidance from different files is detected before rollout. Which implementation should be selected?
- audit overlapping CLAUDE.md files and resolve instruction contradictions
- keep the top-level instruction set concise and move specialized details to scoped files
- assume the model will infer which contradictory rule to follow
- put concise shared guidance in the repository root CLAUDE.md
Explanation
Correct Answer
A:For a deployment-readiness verification process, clear ownership and scope prevent two directions from competing in the same coding task; this directly establishes that conflicting guidance from different files is detected before rollout instead of relying on an unrelated workflow control.
Incorrect Answers
B:Although focused guidance preserves useful context and reduces instruction competition, that only supports the distinct goal that project guidance fits within a limited context budget, not the requirement that conflicting guidance from different files is detected before rollout in a deployment-readiness verification process.
C:In a deployment-readiness verification process, unresolved conflicts create inconsistent output across directories; as a result, the stated requirement that conflicting guidance from different files is detected before rollout would still not be met.
D:Although project instructions travel with the codebase and establish common architecture and build constraints, that only supports the distinct goal that all contributors receive the repository's stable engineering conventions, not the requirement that conflicting guidance from different files is detected before rollout in a deployment-readiness verification process.
Question 16
Tests for a shared-library improvement effort will pass only if the assistant can use external service capabilities via MCP and if sensitive tool calls are denied before they reach an executor. Which proposed design directly satisfies both checks?
- configure a compatible MCP server and expose its declared tools to the client; and register a PreToolUse hook that checks the operation against policy
- describe the external API in a prompt without connecting any tool; and write a stronger instruction in the system prompt
- configure a compatible MCP server and expose its declared tools to the client; and write a stronger instruction in the system prompt
- describe the external API in a prompt without connecting any tool; and register a PreToolUse hook that checks the operation against policy
Explanation
Correct Answer
A:In a shared-library improvement effort, MCP bridges a client to provider-defined capabilities rather than inventing a Claude built-in; alongside this, pre-execution interception can reject a disallowed call rather than relying on post-hoc advice, satisfying the distinct requirements that the assistant can use external service capabilities via MCP and that sensitive tool calls are denied before they reach an executor.
Incorrect Answers
B:In a shared-library improvement effort, instructions cannot create an executable integration endpoint and an instruction alone does not provide deterministic execution-time authorization; the design would fail both conditions: the assistant can use external service capabilities via MCP and sensitive tool calls are denied before they reach an executor.
C:In a shared-library improvement effort, MCP bridges a client to provider-defined capabilities rather than inventing a Claude built-in, but an instruction alone does not provide deterministic execution-time authorization; therefore the separate requirement that sensitive tool calls are denied before they reach an executor would remain unsatisfied.
D:In a shared-library improvement effort, pre-execution interception can reject a disallowed call rather than relying on post-hoc advice, but instructions cannot create an executable integration endpoint; therefore the separate requirement that the assistant can use external service capabilities via MCP would remain unsatisfied.
Question 17
When designing a branch protection review bot, the requirement is not satisfied unless a straightforward lookup avoids unnecessary agent coordination and a developer returns to an interrupted session with prior reasoning. Which implementation addresses the two independent conditions?
- spawn a coordinator plus multiple specialists for every trivial query; and start an unrelated new session and assume it knows past actions
- use one focused tool-assisted agent for the bounded task; and resume the stored session by its identifier
- spawn a coordinator plus multiple specialists for every trivial query; and resume the stored session by its identifier
- use one focused tool-assisted agent for the bounded task; and start an unrelated new session and assume it knows past actions
Explanation
Correct Answer
B:In a branch protection review bot, orchestration overhead is not justified when there is no useful specialization or concurrency; alongside this, session resumption retains the recorded conversation and avoids rebuilding the entire task from scratch, satisfying the distinct requirements that a straightforward lookup avoids unnecessary agent coordination and that a developer returns to an interrupted session with prior reasoning.
Incorrect Answers
A:In a branch protection review bot, extra agents add latency and handoff risk without new capabilities and a fresh context lacks the prior interaction unless the facts are deliberately supplied; the design would fail both conditions: a straightforward lookup avoids unnecessary agent coordination and a developer returns to an interrupted session with prior reasoning.
C:In a branch protection review bot, session resumption retains the recorded conversation and avoids rebuilding the entire task from scratch, but extra agents add latency and handoff risk without new capabilities; therefore the separate requirement that a straightforward lookup avoids unnecessary agent coordination would remain unsatisfied.
D:In a branch protection review bot, orchestration overhead is not justified when there is no useful specialization or concurrency, but a fresh context lacks the prior interaction unless the facts are deliberately supplied; therefore the separate requirement that a developer returns to an interrupted session with prior reasoning would remain unsatisfied.
Question 18
In a service dependency exploration tool, the engineering team must guarantee that a team compares two candidate patches against the same baseline. What is the most appropriate control?
- choose whichever patch has the more persuasive natural-language summary
- evaluate alternatives using the same tests and scoped diffs
- turn the feedback into a concrete test or change request
- iterate with focused diffs rather than wholesale regeneration
Explanation
Correct Answer
B:For a service dependency exploration tool, consistent acceptance evidence reveals which candidate improves the relevant behavior; this directly establishes that a team compares two candidate patches against the same baseline instead of relying on an unrelated workflow control.
Incorrect Answers
A:In a service dependency exploration tool, rhetorical fluency does not establish code correctness; as a result, the stated requirement that a team compares two candidate patches against the same baseline would still not be met.
C:Although actionable feedback narrows the next edit and makes improvement measurable, that only supports the distinct goal that reviewer feedback points to a specific missed requirement, not the requirement that a team compares two candidate patches against the same baseline in a service dependency exploration tool.
D:Although narrow changes are easier to test and evaluate against explicit requirements, that only supports the distinct goal that the review produces a small, auditable patch, not the requirement that a team compares two candidate patches against the same baseline in a service dependency exploration tool.
Question 19
A team operating a shared-library improvement effort wants to establish two separate behaviors: a common security rule remains active across the project; in addition, a delegated reviewer has no production write capability. Which plan implements both?
- keep project-wide security guidance in shared instructions rather than a narrow path glob; and provide only the read-oriented tools required by that reviewer
- keep project-wide security guidance in shared instructions rather than a narrow path glob; and give the reviewer every administrative tool and ask for restraint
- scope the security instruction only to one unrelated test directory; and give the reviewer every administrative tool and ask for restraint
- scope the security instruction only to one unrelated test directory; and provide only the read-oriented tools required by that reviewer
Explanation
Correct Answer
A:In a shared-library improvement effort, cross-cutting obligations should not disappear when an agent changes directories; alongside this, agent-scoped tool exposure prevents a write path in a read-only role, satisfying the distinct requirements that a common security rule remains active across the project and that a delegated reviewer has no production write capability.
Incorrect Answers
B:In a shared-library improvement effort, cross-cutting obligations should not disappear when an agent changes directories, but prompt restraint is weaker than withholding execution capability; therefore the separate requirement that a delegated reviewer has no production write capability would remain unsatisfied.
C:In a shared-library improvement effort, other code paths would lack the universal control and prompt restraint is weaker than withholding execution capability; the design would fail both conditions: a common security rule remains active across the project and a delegated reviewer has no production write capability.
D:In a shared-library improvement effort, agent-scoped tool exposure prevents a write path in a read-only role, but other code paths would lack the universal control; therefore the separate requirement that a common security rule remains active across the project would remain unsatisfied.
Question 20
For a mobile client bug-fix task, a reviewer identifies two independent obligations: separate code modules can be inspected in parallel and tool-call logging does not modify a successful tool result. Which proposal should be approved?
- implement only spawn independent specialists in the same delegation stage; make no corresponding change for the second requirement
- spawn independent specialists in the same delegation stage; and rewrite every tool result as a free-form paraphrase
- make the second specialist wait for the first even though their modules are independent; and rewrite every tool result as a free-form paraphrase
- make the second specialist wait for the first even though their modules are independent; and use an observation-only hook for telemetry when rewriting is unnecessary
- spawn independent specialists in the same delegation stage; and use an observation-only hook for telemetry when rewriting is unnecessary
Explanation
Correct Answer
E:In a mobile client bug-fix task, concurrent subagents can examine non-overlapping modules before the coordinator merges their reports; alongside this, separating instrumentation from mutation reduces the risk of corrupting evidence, satisfying the distinct requirements that separate code modules can be inspected in parallel and that tool-call logging does not modify a successful tool result.
Incorrect Answers
A:In a mobile client bug-fix task, concurrent subagents can examine non-overlapping modules before the coordinator merges their reports, but the independent mechanism for the other criterion is absent: separating instrumentation from mutation reduces the risk of corrupting evidence; as written, the design does not ensure that tool-call logging does not modify a successful tool result.
B:In a mobile client bug-fix task, concurrent subagents can examine non-overlapping modules before the coordinator merges their reports, but summaries can discard fields that subsequent steps require; therefore the separate requirement that tool-call logging does not modify a successful tool result would remain unsatisfied.
C:In a mobile client bug-fix task, serialization wastes time without supplying a dependency and summaries can discard fields that subsequent steps require; the design would fail both conditions: separate code modules can be inspected in parallel and tool-call logging does not modify a successful tool result.
D:In a mobile client bug-fix task, separating instrumentation from mutation reduces the risk of corrupting evidence, but serialization wastes time without supplying a dependency; therefore the separate requirement that separate code modules can be inspected in parallel would remain unsatisfied.