Question 600
The release checklist for a billing-dispute support agent explicitly checks that important workflow state survives context compression or process restarts; it also checks that a broad tool catalog does not overwhelm a small task. What coordinated solution meets the checklist?
- persist authoritative task facts and artifact IDs outside ephemeral conversation text; and supply only context-relevant tools for the current agent
- depend on the model remembering a prior value forever; and load all available MCP tools into every specialist turn
- depend on the model remembering a prior value forever; and supply only context-relevant tools for the current agent
- persist authoritative task facts and artifact IDs outside ephemeral conversation text; and load all available MCP tools into every specialist turn
Explanation
Correct Answer
A:In a billing-dispute support agent, external durable state can be reloaded and checked when the session window changes; alongside this, reducing irrelevant tool definitions improves selection and saves context, satisfying the distinct requirements that important workflow state survives context compression or process restarts and that a broad tool catalog does not overwhelm a small task.
Incorrect Answers
B:In a billing-dispute support agent, context limits and summarization can lose exact identifiers and tool bloat makes the available actions harder to discriminate; the design would fail both conditions: important workflow state survives context compression or process restarts and a broad tool catalog does not overwhelm a small task.
C:In a billing-dispute support agent, reducing irrelevant tool definitions improves selection and saves context, but context limits and summarization can lose exact identifiers; therefore the separate requirement that important workflow state survives context compression or process restarts would remain unsatisfied.
D:In a billing-dispute support agent, external durable state can be reloaded and checked when the session window changes, but tool bloat makes the available actions harder to discriminate; therefore the separate requirement that a broad tool catalog does not overwhelm a small task would remain unsatisfied.
Question 601
An incident postmortem for a CLI application maintenance task requires controls so that a repeatable code-review procedure is available as a named workflow and that a delegated reviewer has no production write capability. Which paired remedy covers both failure modes?
- create a documented reusable skill in SKILL.md; and give the reviewer every administrative tool and ask for restraint
- retype the entire procedure in every conversation; and provide only the read-oriented tools required by that reviewer
- create a documented reusable skill in SKILL.md; and provide only the read-oriented tools required by that reviewer
- retype the entire procedure in every conversation; and give the reviewer every administrative tool and ask for restraint
Explanation
Correct Answer
C:In a CLI application maintenance task, skills package instructions and supporting resources for invocation across similar tasks; alongside this, agent-scoped tool exposure prevents a write path in a read-only role, satisfying the distinct requirements that a repeatable code-review procedure is available as a named workflow and that a delegated reviewer has no production write capability.
Incorrect Answers
A:In a CLI application maintenance task, skills package instructions and supporting resources for invocation across similar tasks, 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.
B:In a CLI application maintenance task, agent-scoped tool exposure prevents a write path in a read-only role, but manual repetition produces drift and loses shared revisions; therefore the separate requirement that a repeatable code-review procedure is available as a named workflow would remain unsatisfied.
D:In a CLI application maintenance task, manual repetition produces drift and loses shared revisions and prompt restraint is weaker than withholding execution capability; the design would fail both conditions: a repeatable code-review procedure is available as a named workflow and a delegated reviewer has no production write capability.
Question 602
Before expanding a static analysis automation, the owner must demonstrate that ambiguous requirements are resolved before a destructive migration, while also demonstrating that project guidance fits within a limited context budget. Which design provides both controls?
- make assumptions and migrate production files immediately; and paste complete design manuals into the top-level CLAUDE.md
- clarify scope and acceptance criteria while planning; and paste complete design manuals into the top-level CLAUDE.md
- clarify scope and acceptance criteria while planning; and keep the top-level instruction set concise and move specialized details to scoped files
- make assumptions and migrate production files immediately; and keep the top-level instruction set concise and move specialized details to scoped files
Explanation
Correct Answer
C:In a static analysis automation, a reviewed plan should resolve unknown targets and rollback expectations before writes; alongside this, focused guidance preserves useful context and reduces instruction competition, satisfying the distinct requirements that ambiguous requirements are resolved before a destructive migration and that project guidance fits within a limited context budget.
Incorrect Answers
A:In a static analysis automation, guessed intent can create costly irreversible changes and indiscriminate loading spends tokens on facts unrelated to the active task; the design would fail both conditions: ambiguous requirements are resolved before a destructive migration and project guidance fits within a limited context budget.
B:In a static analysis automation, a reviewed plan should resolve unknown targets and rollback expectations before writes, but indiscriminate loading spends tokens on facts unrelated to the active task; therefore the separate requirement that project guidance fits within a limited context budget would remain unsatisfied.
D:In a static analysis automation, focused guidance preserves useful context and reduces instruction competition, but guessed intent can create costly irreversible changes; therefore the separate requirement that ambiguous requirements are resolved before a destructive migration would remain unsatisfied.
Question 603
In a technical review of a customs paperwork processor, the architect must address the following constraints together: a destructive tool receives human authorization at execution time; a handoff retains only action-relevant context. Which change set is most defensible?
- collect a general approval at the start of the conversation; and pass a compact structured record of facts, decisions, pending tasks, and sources
- interrupt or deny the tool until an authorization callback approves it; and send every previous chat turn to each new worker
- interrupt or deny the tool until an authorization callback approves it; and pass a compact structured record of facts, decisions, pending tasks, and sources
- collect a general approval at the start of the conversation; and send every previous chat turn to each new worker
Explanation
Correct Answer
C:In a customs paperwork processor, execution gates make the approval binding on the selected action and inputs; alongside this, focused handoffs reduce context size while protecting continuity-critical details, satisfying the distinct requirements that a destructive tool receives human authorization at execution time and that a handoff retains only action-relevant context.
Incorrect Answers
A:In a customs paperwork processor, focused handoffs reduce context size while protecting continuity-critical details, but blanket consent does not verify the later high-risk operation; therefore the separate requirement that a destructive tool receives human authorization at execution time would remain unsatisfied.
B:In a customs paperwork processor, execution gates make the approval binding on the selected action and inputs, but large histories include irrelevant or outdated information; therefore the separate requirement that a handoff retains only action-relevant context would remain unsatisfied.
D:In a customs paperwork processor, blanket consent does not verify the later high-risk operation and large histories include irrelevant or outdated information; the design would fail both conditions: a destructive tool receives human authorization at execution time and a handoff retains only action-relevant context.
Question 604
A team operating a warranty claims assistant wants to establish two separate behaviors: a support case missing an essential fact is not silently guessed; in addition, only one approved write tool is callable in a sensitive stage. Which plan implements both?
- ask a focused clarifying question for the missing requirement; and restrict the stage's allowed tool set and verify permissions in the executor
- invent a plausible value and continue automatically; and let the model select from a broad global list of tools
- invent a plausible value and continue automatically; and restrict the stage's allowed tool set and verify permissions in the executor
- ask a focused clarifying question for the missing requirement; and let the model select from a broad global list of tools
Explanation
Correct Answer
A:In a warranty claims assistant, targeted clarification supplies the decision input before selecting an irreversible action; alongside this, capability scoping narrows possible side effects before model selection, satisfying the distinct requirements that a support case missing an essential fact is not silently guessed and that only one approved write tool is callable in a sensitive stage.
Incorrect Answers
B:In a warranty claims assistant, fabricated facts can cause the wrong account or policy to be applied and excess capabilities create avoidable routing and authorization risk; the design would fail both conditions: a support case missing an essential fact is not silently guessed and only one approved write tool is callable in a sensitive stage.
C:In a warranty claims assistant, capability scoping narrows possible side effects before model selection, but fabricated facts can cause the wrong account or policy to be applied; therefore the separate requirement that a support case missing an essential fact is not silently guessed would remain unsatisfied.
D:In a warranty claims assistant, targeted clarification supplies the decision input before selecting an irreversible action, but excess capabilities create avoidable routing and authorization risk; therefore the separate requirement that only one approved write tool is callable in a sensitive stage would remain unsatisfied.
Question 605
The release checklist for a continuous integration repair assistant explicitly checks that agents exchange outputs without losing important structure; it also checks that sensitive tool calls are denied before they reach an executor. What coordinated solution meets the checklist?
- define artifact schemas and handoff contracts between roles; and register a PreToolUse hook that checks the operation against policy
- pass raw chat logs between every stage; and register a PreToolUse hook that checks the operation against policy
- pass raw chat logs between every stage; and write a stronger instruction in the system prompt
- define artifact schemas and handoff contracts between roles; and write a stronger instruction in the system prompt
Explanation
Correct Answer
A:In a continuous integration repair assistant, bounded interfaces make missing fields and incompatible assumptions visible; alongside this, pre-execution interception can reject a disallowed call rather than relying on post-hoc advice, satisfying the distinct requirements that agents exchange outputs without losing important structure and that sensitive tool calls are denied before they reach an executor.
Incorrect Answers
B:In a continuous integration repair assistant, pre-execution interception can reject a disallowed call rather than relying on post-hoc advice, but unstructured context is noisy and can obscure the decision fields; therefore the separate requirement that agents exchange outputs without losing important structure would remain unsatisfied.
C:In a continuous integration repair assistant, unstructured context is noisy and can obscure the decision fields and an instruction alone does not provide deterministic execution-time authorization; the design would fail both conditions: agents exchange outputs without losing important structure and sensitive tool calls are denied before they reach an executor.
D:In a continuous integration repair assistant, bounded interfaces make missing fields and incompatible assumptions visible, 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.
Question 606
For a source-navigation tool for new hires, a reviewer identifies two independent obligations: project guidance fits within a limited context budget and a paused server-side tool loop can continue without losing work. Which proposal should be approved?
- keep the top-level instruction set concise and move specialized details to scoped files; and preserve the paused assistant response and resume the server-tool turn
- paste complete design manuals into the top-level CLAUDE.md; and restart from a blank conversation whenever pause_turn occurs
- keep the top-level instruction set concise and move specialized details to scoped files; and restart from a blank conversation whenever pause_turn occurs
- paste complete design manuals into the top-level CLAUDE.md; and preserve the paused assistant response and resume the server-tool turn
Explanation
Correct Answer
A:In a source-navigation tool for new hires, focused guidance preserves useful context and reduces instruction competition; alongside this, pause_turn indicates an internal tool loop pause rather than a user-facing conclusion, satisfying the distinct requirements that project guidance fits within a limited context budget and that a paused server-side tool loop can continue without losing work.
Incorrect Answers
B:In a source-navigation tool for new hires, indiscriminate loading spends tokens on facts unrelated to the active task and dropping tool context can repeat work and omit intermediate evidence; the design would fail both conditions: project guidance fits within a limited context budget and a paused server-side tool loop can continue without losing work.
C:In a source-navigation tool for new hires, focused guidance preserves useful context and reduces instruction competition, but dropping tool context can repeat work and omit intermediate evidence; therefore the separate requirement that a paused server-side tool loop can continue without losing work would remain unsatisfied.
D:In a source-navigation tool for new hires, pause_turn indicates an internal tool loop pause rather than a user-facing conclusion, but indiscriminate loading spends tokens on facts unrelated to the active task; therefore the separate requirement that project guidance fits within a limited context budget would remain unsatisfied.
Question 607
When designing a returns-policy assistant, the requirement is not satisfied unless the retry process does not create duplicate records and a failed retrieval agent does not masquerade as having found no matches. Which implementation addresses the two independent conditions?
- insert a fresh record on every model retry; and return a typed failure status with its origin to the coordinator
- assign stable source identifiers and deduplicate writes; and convert every retrieval failure into an empty list
- assign stable source identifiers and deduplicate writes; and return a typed failure status with its origin to the coordinator
- insert a fresh record on every model retry; and convert every retrieval failure into an empty list
Explanation
Correct Answer
C:In a returns-policy assistant, idempotent processing separates repeated model attempts from committed business mutations; alongside this, an operational error must remain distinguishable from a valid empty result, satisfying the distinct requirements that the retry process does not create duplicate records and that a failed retrieval agent does not masquerade as having found no matches.
Incorrect Answers
A:In a returns-policy assistant, an operational error must remain distinguishable from a valid empty result, but repeated extraction attempts could create duplicate downstream state; therefore the separate requirement that the retry process does not create duplicate records would remain unsatisfied.
B:In a returns-policy assistant, idempotent processing separates repeated model attempts from committed business mutations, but the final report could incorrectly claim no sources exist; therefore the separate requirement that a failed retrieval agent does not masquerade as having found no matches would remain unsatisfied.
D:In a returns-policy assistant, repeated extraction attempts could create duplicate downstream state and the final report could incorrectly claim no sources exist; the design would fail both conditions: the retry process does not create duplicate records and a failed retrieval agent does not masquerade as having found no matches.
Question 608
In an analyst research coordinator, the acceptance criteria require that near-overlapping retrieval tools are selected by actual boundary and that a broad tool catalog does not overwhelm a small task. Which paired implementation meets both?
- document examples and conditions where each tool must not be used; and load all available MCP tools into every specialist turn
- document examples and conditions where each tool must not be used; and supply only context-relevant tools for the current agent
- rely exclusively on short tool names for routing; and supply only context-relevant tools for the current agent
- rely exclusively on short tool names for routing; and load all available MCP tools into every specialist turn
Explanation
Correct Answer
B:In an analyst research coordinator, contrastive guidance helps tool choice when resource names look similar; alongside this, reducing irrelevant tool definitions improves selection and saves context, satisfying the distinct requirements that near-overlapping retrieval tools are selected by actual boundary and that a broad tool catalog does not overwhelm a small task.
Incorrect Answers
A:In an analyst research coordinator, contrastive guidance helps tool choice when resource names look similar, but tool bloat makes the available actions harder to discriminate; therefore the separate requirement that a broad tool catalog does not overwhelm a small task would remain unsatisfied.
C:In an analyst research coordinator, reducing irrelevant tool definitions improves selection and saves context, but names alone may not capture the distinction in scope or authorization; therefore the separate requirement that near-overlapping retrieval tools are selected by actual boundary would remain unsatisfied.
D:In an analyst research coordinator, names alone may not capture the distinction in scope or authorization and tool bloat makes the available actions harder to discriminate; the design would fail both conditions: near-overlapping retrieval tools are selected by actual boundary and a broad tool catalog does not overwhelm a small task.
Question 609
While preparing a research PDF metadata pipeline for deployment, engineers must ensure that an MCP tool failure can be traced across systems; they must also ensure that a delegated reviewer has no production write capability. Which combination best satisfies those requirements?
- replace every failure with the same unstructured unknown error; and give the reviewer every administrative tool and ask for restraint
- implement only provide only the read-oriented tools required by that reviewer; make no corresponding change for the first requirement
- propagate a stable correlation ID and machine-readable error code; and provide only the read-oriented tools required by that reviewer
- replace every failure with the same unstructured unknown error; and provide only the read-oriented tools required by that reviewer
- implement only propagate a stable correlation ID and machine-readable error code; make no corresponding change for the second requirement
- propagate a stable correlation ID and machine-readable error code; and give the reviewer every administrative tool and ask for restraint
Explanation
Correct Answer
C:In a research PDF metadata pipeline, cross-service identifiers connect the user's symptom to upstream logs without exposing secrets; alongside this, agent-scoped tool exposure prevents a write path in a read-only role, satisfying the distinct requirements that an MCP tool failure can be traced across systems and that a delegated reviewer has no production write capability.
Incorrect Answers
A:In a research PDF metadata pipeline, undifferentiated errors prevent meaningful triage and prompt restraint is weaker than withholding execution capability; the design would fail both conditions: an MCP tool failure can be traced across systems and a delegated reviewer has no production write capability.
B:In a research PDF metadata pipeline, agent-scoped tool exposure prevents a write path in a read-only role, but the independent mechanism for the other criterion is absent: cross-service identifiers connect the user's symptom to upstream logs without exposing secrets; as written, the design does not ensure that an MCP tool failure can be traced across systems.
D:In a research PDF metadata pipeline, agent-scoped tool exposure prevents a write path in a read-only role, but undifferentiated errors prevent meaningful triage; therefore the separate requirement that an MCP tool failure can be traced across systems would remain unsatisfied.
E:In a research PDF metadata pipeline, cross-service identifiers connect the user's symptom to upstream logs without exposing secrets, but the independent mechanism for the other criterion is absent: agent-scoped tool exposure prevents a write path in a read-only role; as written, the design does not ensure that a delegated reviewer has no production write capability.
F:In a research PDF metadata pipeline, cross-service identifiers connect the user's symptom to upstream logs without exposing secrets, 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.
Question 610
The release checklist for a customer case-management agent explicitly checks that out-of-policy exceptions receive an authorized decision; it also checks that each tool interface has a single clear operation boundary. What coordinated solution meets the checklist?
- let the agent approve an exception whenever its wording sounds confident; and separate read-only lookup from side-effecting update interfaces
- route the exception to a human approval step before execution; and use a single universal tool that infers read versus write from prose
- let the agent approve an exception whenever its wording sounds confident; and use a single universal tool that infers read versus write from prose
- route the exception to a human approval step before execution; and separate read-only lookup from side-effecting update interfaces
Explanation
Correct Answer
D:In a customer case-management agent, approval is a workflow boundary separate from language-model confidence; alongside this, clear operation boundaries support least privilege and understandable approvals, satisfying the distinct requirements that out-of-policy exceptions receive an authorized decision and that each tool interface has a single clear operation boundary.
Incorrect Answers
A:In a customer case-management agent, clear operation boundaries support least privilege and understandable approvals, but model confidence is not delegated business authority; therefore the separate requirement that out-of-policy exceptions receive an authorized decision would remain unsatisfied.
B:In a customer case-management agent, approval is a workflow boundary separate from language-model confidence, 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.
C:In a customer case-management agent, model confidence is not delegated business authority and overloading hides effects and makes execution review harder; the design would fail both conditions: out-of-policy exceptions receive an authorized decision and each tool interface has a single clear operation boundary.