Question 680
The acceptance criteria for a workflow automation change state that it must select orchestration dependencies for a project that already has a stable task engine. Which implementation aligns with that criterion?
- Prefer a finite state machine for regulated steps
- Use a plan-execute-review loop
- Use a graceful fallback path
- Use a framework only when its primitives fit
- Delegate narrow subtasks with a separate context
Explanation
Correct Answer
D:Frameworks can help with graphs, retries, and checkpoints but create maintenance and integration costs; this behavior directly supports the need to select orchestration dependencies for a project that already has a stable task engine. Adopting a framework solely for fashionable abstraction is not a technical requirement.
Incorrect Answers
A:Explicit states and transitions control who may advance a case independently of generated language; the supported outcome is to enforce approval, execution, and confirmation order in a compliance workflow, whereas this scenario requires the ability to select orchestration dependencies for a project that already has a stable task engine; an unrestricted agent loop can skip expected governance stages.
B:This approach serves the separate need to manage a complex migration that requires checkpoints after several tool actions because separating plan, execution, and review makes deviations visible before a final outcome is accepted; it does not establish the outcome to select orchestration dependencies for a project that already has a stable task engine, and a one-shot response cannot verify environmental effects.
C:A bounded failure branch can return a safe partial result or route to a human rather than loop indefinitely; despite being useful when a service must keep a customer service application operating when an expensive specialist agent cannot complete, it is not a solution to select orchestration dependencies for a project that already has a stable task engine; unbounded retries do not ensure eventual success.
E:Each subagent can work in a bounded context and return a concise, verified artifact; this applies to the different goal to protect a coordinator's window from huge specialist tool outputs, not the requirement to select orchestration dependencies for a project that already has a stable task engine; copying every intermediate trace into the parent inflates context and cost.
Question 681
Before a prompt engineering release, three separately owned tasks must be covered: one service must prevent a customer-facing assistant from making final legal determinations without review; another must avoid invented contract dates when the uploaded document lacks them; and a third must generate a technical incident summary suitable for an on-call engineer instead of an executive. Which THREE choices directly address the stated tasks? Choose THREE.
- Constrain response audience and purpose
- Specify the handling of missing information
- Specify output format in the prompt
- Delimit untrusted source material
- Use explicit refusal and escalation instructions
- Separate system policy from user content
Explanation
Correct Answers
A:Audience constraints determine useful terminology, specificity, and level of detail; this directly supports task 3 to generate a technical incident summary suitable for an on-call engineer instead of an executive, not the separate task or tasks to prevent a customer-facing assistant from making final legal determinations without review and to avoid invented contract dates when the uploaded document lacks them; style guidance alone cannot make inaccurate facts true.
B:An explicit unknown or null rule makes uncertainty visible in the requested output; this directly supports task 2 to avoid invented contract dates when the uploaded document lacks them, not the separate task or tasks to prevent a customer-facing assistant from making final legal determinations without review and to generate a technical incident summary suitable for an on-call engineer instead of an executive; a strict field list is not permission to fabricate absent values.
E:Clear boundaries describe when to abstain or route to a qualified person; this directly supports task 1 to prevent a customer-facing assistant from making final legal determinations without review, not the separate task or tasks to avoid invented contract dates when the uploaded document lacks them and to generate a technical incident summary suitable for an on-call engineer instead of an executive; text instructions require supporting application enforcement for high-risk decisions.
Incorrect Answers
C:Explicit field descriptions guide generation toward the consumer's expected structure; this implements the different outcome to obtain named fields in a consistent response layout for subsequent processing rather than the required tasks to generate a technical incident summary suitable for an on-call engineer instead of an executive and to prevent a customer-facing assistant from making final legal determinations without review and to avoid invented contract dates when the uploaded document lacks them, and formatting guidance still requires machine validation at the boundary, so it is not a direct solution to any of the assigned workstreams.
D:Clear source boundaries and explicit treatment as quoted evidence help reduce instruction confusion; this implements the different outcome to let the model summarize a hostile email without obeying its embedded commands rather than the required tasks to generate a technical incident summary suitable for an on-call engineer instead of an executive and to prevent a customer-facing assistant from making final legal determinations without review and to avoid invented contract dates when the uploaded document lacks them, and delimiters alone cannot replace permission checks on dangerous tools, so it is not a direct solution to any of the assigned workstreams.
F:Stable higher-priority instructions should be represented in the designated instruction channel; this implements the different outcome to prevent a customer message from overriding the application's escalation rules rather than the required tasks to generate a technical incident summary suitable for an on-call engineer instead of an executive and to prevent a customer-facing assistant from making final legal determinations without review and to avoid invented contract dates when the uploaded document lacks them, and user text is the task data, not a privileged configuration surface, so it is not a direct solution to any of the assigned workstreams.
Question 682
An external system access review requires two independent, explicitly defined outcomes: the first workstream must automate the same classify-review-export path across predictable submissions; a second must deny destructive filesystem operations before a coding agent executes them. Which TWO actions directly provide these capabilities? Choose TWO.
- Use hooks for enforceable safety checks
- Use system or repository instructions for stable behavior
- Use subagents for independent expertise
- Use a deterministic workflow for fixed stages
- Use an autonomous agent for exploratory work
Explanation
Correct Answers
A:Runtime interception can apply deterministic policies to proposed actions; this directly supports task 2 to deny destructive filesystem operations before a coding agent executes them, not the separate task or tasks to automate the same classify-review-export path across predictable submissions; a skill that asks for caution does not guarantee prevention.
D:Explicit application control makes each transition testable and easier to audit; this directly supports task 1 to automate the same classify-review-export path across predictable submissions, not the separate task or tasks to deny destructive filesystem operations before a coding agent executes them; an autonomous agent is unnecessary when the path is already fixed.
Incorrect Answers
B:Instructions shape model behavior and task framing without executing external operations; this implements the different outcome to make a coding or API assistant follow organization-wide conventions rather than the required tasks to automate the same classify-review-export path across predictable submissions and to deny destructive filesystem operations before a coding agent executes them, and descriptive rules do not replace programmatic validation or IAM, so it is not a direct solution to any of the assigned workstreams.
C:Specialist contexts reduce tool-output contamination and support focused parallel work; this implements the different outcome to delegate a security analysis while keeping the main coordinator's context concise rather than the required tasks to automate the same classify-review-export path across predictable submissions and to deny destructive filesystem operations before a coding agent executes them, and a reusable Skill alone does not create an independent worker with separate state, so it is not a direct solution to any of the assigned workstreams.
E:Model-directed planning supports adaptive sequences when the next action is not known in advance; this implements the different outcome to perform open-ended troubleshooting where tool selection changes after each observation rather than the required tasks to automate the same classify-review-export path across predictable submissions and to deny destructive filesystem operations before a coding agent executes them, and a rigid workflow struggles with unforeseen branches without extensive coding, so it is not a direct solution to any of the assigned workstreams.
Question 683
During a backend integration rollout, two different failure paths need remediation. The first component must distinguish an invalid request from a temporary provider or rate-limit failure; separately, the second must record complete streamed output and token usage accurately. Which changes should be adopted?
- First component: Read the HTTP error status before retrying. Second component: Handle the tool_use stop_reason.
- First component: Handle the tool_use stop_reason. Second component: Process tool_use responses in the application.
- First component: Read the HTTP error status before retrying. Second component: Handle message_stop and final usage events.
- First component: Process tool_use responses in the application. Second component: Handle message_stop and final usage events.
Explanation
Correct Answer
C:The first proposal fits its assigned role because status codes and structured errors guide correction versus bounded retry, while the second proposal fits its assigned role because the final stream events signal completion and deliver accumulated metadata beyond individual text fragments; their respective acceptance criteria are to distinguish an invalid request from a temporary provider or rate-limit failure and to record complete streamed output and token usage accurately.
Incorrect Answers
A:The first proposal fits its assigned role because status codes and structured errors guide correction versus bounded retry, while the second proposal serves the different purpose to continue the agent loop when Claude is waiting for external observations instead of the assigned second role because this stop reason means tool calls need to be executed and their results returned before synthesis; their respective acceptance criteria are to distinguish an invalid request from a temporary provider or rate-limit failure and to record complete streamed output and token usage accurately.
B:The first proposal serves the different purpose to continue the agent loop when Claude is waiting for external observations instead of the assigned first role because this stop reason means tool calls need to be executed and their results returned before synthesis, while the second proposal serves the different purpose to actually retrieve a customer's order status requested by Claude instead of the assigned second role because client tools execute in application code; the model emits structured requests but does not invoke external functions itself; their respective acceptance criteria are to distinguish an invalid request from a temporary provider or rate-limit failure and to record complete streamed output and token usage accurately.
D:The first proposal serves the different purpose to actually retrieve a customer's order status requested by Claude instead of the assigned first role because client tools execute in application code; the model emits structured requests but does not invoke external functions itself, while the second proposal fits its assigned role because the final stream events signal completion and deliver accumulated metadata beyond individual text fragments; their respective acceptance criteria are to distinguish an invalid request from a temporary provider or rate-limit failure and to record complete streamed output and token usage accurately.
Question 684
After pilot feedback, an inference routing group needs two independent improvements. The first is to call the Messages endpoint from a language without a mature provider SDK; the second is to deliver tokens to a browser without maintaining a bidirectional command channel. Which engineering proposal handles both?
- First component: Use a queue for burst absorption. Second component: Choose SSE for one-way streamed responses.
- First component: Use direct HTTP REST integration for unsupported runtimes. Second component: Use connection pooling for repeated HTTP calls.
- First component: Use direct HTTP REST integration for unsupported runtimes. Second component: Choose SSE for one-way streamed responses.
- First component: Use connection pooling for repeated HTTP calls. Second component: Use a queue for burst absorption.
Explanation
Correct Answer
C:The first proposal fits its assigned role because a standards-based HTTP client can send JSON requests with the correct headers and decode responses, while the second proposal fits its assigned role because server-sent events provide ordered server-to-client updates over an HTTP stream; their respective acceptance criteria are to call the Messages endpoint from a language without a mature provider SDK and to deliver tokens to a browser without maintaining a bidirectional command channel.
Incorrect Answers
A:The first proposal serves the different purpose to smooth sudden ingestion spikes before workers call a capacity-limited model API instead of the assigned first role because queueing decouples producers from consumer throughput and enables controlled retries, while the second proposal fits its assigned role because server-sent events provide ordered server-to-client updates over an HTTP stream; their respective acceptance criteria are to call the Messages endpoint from a language without a mature provider SDK and to deliver tokens to a browser without maintaining a bidirectional command channel.
B:The first proposal fits its assigned role because a standards-based HTTP client can send JSON requests with the correct headers and decode responses, while the second proposal serves the different purpose to reduce setup overhead when a service sends many short requests instead of the assigned second role because reusable clients avoid unnecessary connection establishment under compatible networking conditions; their respective acceptance criteria are to call the Messages endpoint from a language without a mature provider SDK and to deliver tokens to a browser without maintaining a bidirectional command channel.
D:The first proposal serves the different purpose to reduce setup overhead when a service sends many short requests instead of the assigned first role because reusable clients avoid unnecessary connection establishment under compatible networking conditions, while the second proposal serves the different purpose to smooth sudden ingestion spikes before workers call a capacity-limited model API instead of the assigned second role because queueing decouples producers from consumer throughput and enables controlled retries; their respective acceptance criteria are to call the Messages endpoint from a language without a mature provider SDK and to deliver tokens to a browser without maintaining a bidirectional command channel.
Question 685
An operations orchestration team is evaluating two disjoint use cases. For one use case, it must ensure a generated specification is validated before code generation begins; for the other, it must allow a research worker to read records without granting its coordinator's write privileges. Which paired choices are justified?
- First component: Use sequential chaining for dependent subtasks. Second component: Use an explicit completion condition and step budget.
- First component: Use an explicit completion condition and step budget. Second component: Use a deterministic workflow.
- First component: Use sequential chaining for dependent subtasks. Second component: Restrict each specialist's tool permissions.
- First component: Use a deterministic workflow. Second component: Restrict each specialist's tool permissions.
Explanation
Correct Answer
C:The first proposal fits its assigned role because each stage consumes an approved artifact from the previous stage, preserving dependency order and stage-specific checks, while the second proposal fits its assigned role because scoped tool capability prevents a delegated role from using authority unrelated to its assignment; their respective acceptance criteria are to ensure a generated specification is validated before code generation begins and to allow a research worker to read records without granting its coordinator's write privileges.
Incorrect Answers
A:The first proposal fits its assigned role because each stage consumes an approved artifact from the previous stage, preserving dependency order and stage-specific checks, while the second proposal serves the different purpose to bound the runtime of an agent that repeatedly searches without converging instead of the assigned second role because a maximum iteration or cost budget plus a clear success criterion stops runaway tool loops; their respective acceptance criteria are to ensure a generated specification is validated before code generation begins and to allow a research worker to read records without granting its coordinator's write privileges.
B:The first proposal serves the different purpose to bound the runtime of an agent that repeatedly searches without converging instead of the assigned first role because a maximum iteration or cost budget plus a clear success criterion stops runaway tool loops, while the second proposal serves the different purpose to execute an unchanged sequence of extraction, validation, and publishing steps instead of the assigned second role because a fixed pipeline keeps control flow in application code, making execution order reproducible and easy to audit; their respective acceptance criteria are to ensure a generated specification is validated before code generation begins and to allow a research worker to read records without granting its coordinator's write privileges.
D:The first proposal serves the different purpose to execute an unchanged sequence of extraction, validation, and publishing steps instead of the assigned first role because a fixed pipeline keeps control flow in application code, making execution order reproducible and easy to audit, while the second proposal fits its assigned role because scoped tool capability prevents a delegated role from using authority unrelated to its assignment; their respective acceptance criteria are to ensure a generated specification is validated before code generation begins and to allow a research worker to read records without granting its coordinator's write privileges.
Question 686
An application security review requires two independent, explicitly defined outcomes: the first workstream must restore a safe service state after a new guardrail configuration causes widespread errors; a second must let a support assistant read shipment status without issuing refunds. Which TWO actions directly provide these capabilities? Choose TWO.
- Sandbox generated code execution
- Use layered checks for unsafe requests
- Build a documented rollback and incident process
- Run adversarial prompt-injection tests
- Enforce least-privilege tool permissions
Explanation
Correct Answers
C:Versioned release artifacts and response runbooks support containment and recovery; this directly supports task 1 to restore a safe service state after a new guardrail configuration causes widespread errors, not the separate task or tasks to let a support assistant read shipment status without issuing refunds; improvised rollback during an outage increases operational risk.
E:Restricted scopes and read-only interfaces limit impact if a tool request is manipulated; this directly supports task 2 to let a support assistant read shipment status without issuing refunds, not the separate task or tasks to restore a safe service state after a new guardrail configuration causes widespread errors; role language in the prompt cannot prevent an overprivileged API token from writing.
Incorrect Answers
A:Resource, filesystem, and network isolation contain failures and malicious effects; this implements the different outcome to inspect proposed scripts without granting them production host privileges rather than the required tasks to let a support assistant read shipment status without issuing refunds and to restore a safe service state after a new guardrail configuration causes widespread errors, and relying on model intent does not constrain a subprocess, so it is not a direct solution to any of the assigned workstreams.
B:Independent controls catch failures that a single prompt-level rule may miss; this implements the different outcome to detect harmful instructions at input, tool-call, and output boundaries rather than the required tasks to let a support assistant read shipment status without issuing refunds and to restore a safe service state after a new guardrail configuration causes widespread errors, and a system prompt alone is not a complete enforcement strategy, so it is not a direct solution to any of the assigned workstreams.
D:Purposeful hostile test cases probe whether the trust boundary holds under manipulated evidence; this implements the different outcome to find vulnerabilities in a retrieval-augmented assistant before it accesses production tools rather than the required tasks to let a support assistant read shipment status without issuing refunds and to restore a safe service state after a new guardrail configuration causes widespread errors, and ordinary happy-path tests rarely exercise attacker instructions, so it is not a direct solution to any of the assigned workstreams.
Question 687
A design meeting for a document processing platform concludes that two isolated services require different behavior: one must ensure an API assistant consistently uses a defined support role across requests, and another must keep follow-up requests linked without exposing another tenant's messages. Which arrangement meets the specification?
- First component: Limit retained context to authorized history. Second component: Create one session per user conversation.
- First component: Place durable behavior instructions in the system parameter. Second component: Isolate tools behind a service layer.
- First component: Isolate tools behind a service layer. Second component: Limit retained context to authorized history.
- First component: Place durable behavior instructions in the system parameter. Second component: Create one session per user conversation.
Explanation
Correct Answer
D:The first proposal fits its assigned role because system-level guidance supplies application-wide instruction context, separate from each user's task content, while the second proposal fits its assigned role because session scoping ensures history and tool results belong to the correct principal; their respective acceptance criteria are to ensure an API assistant consistently uses a defined support role across requests and to keep follow-up requests linked without exposing another tenant's messages.
Incorrect Answers
A:The first proposal serves the different purpose to prevent a new user from inheriting another user's document excerpts instead of the assigned first role because context assembly must apply access checks before adding prior turns or retrieved evidence, while the second proposal fits its assigned role because session scoping ensures history and tool results belong to the correct principal; their respective acceptance criteria are to ensure an API assistant consistently uses a defined support role across requests and to keep follow-up requests linked without exposing another tenant's messages.
B:The first proposal fits its assigned role because system-level guidance supplies application-wide instruction context, separate from each user's task content, while the second proposal serves the different purpose to avoid exposing raw production database credentials directly to an agent instead of the assigned second role because a mediation service validates identity, arguments, authorization, and audit logs for each requested action; their respective acceptance criteria are to ensure an API assistant consistently uses a defined support role across requests and to keep follow-up requests linked without exposing another tenant's messages.
C:The first proposal serves the different purpose to avoid exposing raw production database credentials directly to an agent instead of the assigned first role because a mediation service validates identity, arguments, authorization, and audit logs for each requested action, while the second proposal serves the different purpose to prevent a new user from inheriting another user's document excerpts instead of the assigned second role because context assembly must apply access checks before adding prior turns or retrieved evidence; their respective acceptance criteria are to ensure an API assistant consistently uses a defined support role across requests and to keep follow-up requests linked without exposing another tenant's messages.
Question 688
Engineers are splitting a data pipeline service into two independently tested functions. Function A must record complete streamed output and token usage accurately, while function B must recover a response that ended because its output budget was exhausted. Which plan is sound?
- First component: Handle message_stop and final usage events. Second component: Use SSE streaming for partial response delivery.
- First component: Submit a Message Batch for offline jobs. Second component: Handle the max_tokens stop_reason.
- First component: Handle message_stop and final usage events. Second component: Handle the max_tokens stop_reason.
- First component: Use SSE streaming for partial response delivery. Second component: Submit a Message Batch for offline jobs.
Explanation
Correct Answer
C:The first proposal fits its assigned role because the final stream events signal completion and deliver accumulated metadata beyond individual text fragments, while the second proposal fits its assigned role because the finish reason identifies truncation, so a caller may need a continuation or a revised token budget; their respective acceptance criteria are to record complete streamed output and token usage accurately and to recover a response that ended because its output budget was exhausted.
Incorrect Answers
A:The first proposal fits its assigned role because the final stream events signal completion and deliver accumulated metadata beyond individual text fragments, while the second proposal serves the different purpose to display a long report as it is generated rather than waiting for the full completion instead of the assigned second role because streaming yields ordered events and content deltas that the client assembles into the final response; their respective acceptance criteria are to record complete streamed output and token usage accurately and to recover a response that ended because its output budget was exhausted.
B:The first proposal serves the different purpose to process thousands of nightly documents without interactive deadlines instead of the assigned first role because batches suit asynchronous throughput-oriented workloads, with results retrieved after processing, while the second proposal fits its assigned role because the finish reason identifies truncation, so a caller may need a continuation or a revised token budget; their respective acceptance criteria are to record complete streamed output and token usage accurately and to recover a response that ended because its output budget was exhausted.
D:The first proposal serves the different purpose to display a long report as it is generated rather than waiting for the full completion instead of the assigned first role because streaming yields ordered events and content deltas that the client assembles into the final response, while the second proposal serves the different purpose to process thousands of nightly documents without interactive deadlines instead of the assigned second role because batches suit asynchronous throughput-oriented workloads, with results retrieved after processing; their respective acceptance criteria are to record complete streamed output and token usage accurately and to recover a response that ended because its output budget was exhausted.
Question 689
The acceptance criteria for a model platform change state that it must spend high-capability tokens only when validation indicates complexity. Which implementation aligns with that criterion?
- Use batch pricing where deadline permits
- Use model routing for simple versus hard cases
- Keep volatile fields after reusable context
- Set budget alerts and tenant caps
- Set maximum step counts for agents
Explanation
Correct Answer
B:Conditional routing improves blended cost without discarding an escalation path; this behavior directly supports the need to spend high-capability tokens only when validation indicates complexity. Blindly downgrading every call can increase retries and review expenses.
Incorrect Answers
A:Asynchronous batch processing can offer a cost-effective throughput path under supported terms; the supported outcome is to lower spend for large scheduled classification jobs that do not need immediate responses, whereas this scenario requires the ability to spend high-capability tokens only when validation indicates complexity; a night job should not be optimized with live streaming unless its users need it.
C:Stable instruction and reference sections should precede changing per-request details for effective prefix reuse; despite being useful when a service must prevent request timestamps from invalidating a costly static prefix, it is not a solution to spend high-capability tokens only when validation indicates complexity; placing a changing value before the cache boundary undermines matching.
D:For the alternate objective to stop a single customer from consuming the entire API allowance during a runaway workflow, accounting plus enforced limits bounds exposure to unexpected traffic and agent loops; the feature here must spend high-capability tokens only when validation indicates complexity, and usage dashboards alone only report overspend after the fact.
E:A bound on iterations prevents repeated generation and calls from growing without limit; this applies to the different goal to contain cumulative token and tool cost across a recursive research session, not the requirement to spend high-capability tokens only when validation indicates complexity; a per-message output cap alone does not control total multi-step spend.
Question 690
Engineers are splitting a retrieval pipeline service into two independently tested functions. Function A must stop a summarizer from producing unrelated analysis instead of required action items, while function B must generate a technical incident summary suitable for an on-call engineer instead of an executive. Which plan is sound?
- First component: Delimit untrusted source material. Second component: Constrain response audience and purpose.
- First component: State one measurable outcome per instruction. Second component: Constrain response audience and purpose.
- First component: Use explicit refusal and escalation instructions. Second component: Delimit untrusted source material.
- First component: State one measurable outcome per instruction. Second component: Use explicit refusal and escalation instructions.
Explanation
Correct Answer
B:The first proposal fits its assigned role because clear task goals and acceptance criteria reduce competing interpretations, while the second proposal fits its assigned role because audience constraints determine useful terminology, specificity, and level of detail; their respective acceptance criteria are to stop a summarizer from producing unrelated analysis instead of required action items and to generate a technical incident summary suitable for an on-call engineer instead of an executive.
Incorrect Answers
A:The first proposal serves the different purpose to let the model summarize a hostile email without obeying its embedded commands instead of the assigned first role because clear source boundaries and explicit treatment as quoted evidence help reduce instruction confusion, while the second proposal fits its assigned role because audience constraints determine useful terminology, specificity, and level of detail; their respective acceptance criteria are to stop a summarizer from producing unrelated analysis instead of required action items and to generate a technical incident summary suitable for an on-call engineer instead of an executive.
C:The first proposal serves the different purpose to prevent a customer-facing assistant from making final legal determinations without review instead of the assigned first role because clear boundaries describe when to abstain or route to a qualified person, while the second proposal serves the different purpose to let the model summarize a hostile email without obeying its embedded commands instead of the assigned second role because clear source boundaries and explicit treatment as quoted evidence help reduce instruction confusion; their respective acceptance criteria are to stop a summarizer from producing unrelated analysis instead of required action items and to generate a technical incident summary suitable for an on-call engineer instead of an executive.
D:The first proposal fits its assigned role because clear task goals and acceptance criteria reduce competing interpretations, while the second proposal serves the different purpose to prevent a customer-facing assistant from making final legal determinations without review instead of the assigned second role because clear boundaries describe when to abstain or route to a qualified person; their respective acceptance criteria are to stop a summarizer from producing unrelated analysis instead of required action items and to generate a technical incident summary suitable for an on-call engineer instead of an executive.