Question 130
An external system access review requires two independent, explicitly defined outcomes: the first workstream must enable a customer service agent to prepare but not autonomously approve account closures; a second must call a protected internal pricing API under the application's own credentials. Which TWO actions directly provide these capabilities? Choose TWO.
- Combine tool access with human confirmation
- Choose a built-in provider tool
- Create a custom client-side tool
- Use system or repository instructions for stable behavior
- Use an autonomous agent for exploratory work
Explanation
Correct Answers
A:Tools supply actionable capabilities while a separate approval step governs sensitive execution; this directly supports task 1 to enable a customer service agent to prepare but not autonomously approve account closures, not the separate task or tasks to call a protected internal pricing API under the application's own credentials; access to a function does not imply authority to invoke it without review.
C:Local execution gives the developer direct control over authentication, validation, and side effects; this directly supports task 2 to call a protected internal pricing API under the application's own credentials, not the separate task or tasks to enable a customer service agent to prepare but not autonomously approve account closures; a hosted generic search feature cannot access private endpoints automatically.
Incorrect Answers
B:Provider-managed execution can reduce implementation work when its scope matches the requirement; this implements the different outcome to use an already supported capability without hosting a duplicate custom service rather than the required tasks to call a protected internal pricing API under the application's own credentials and to enable a customer service agent to prepare but not autonomously approve account closures, and built-in tools may not expose a proprietary enterprise data source, so it is not a direct solution to any of the assigned workstreams.
D: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 call a protected internal pricing API under the application's own credentials and to enable a customer service agent to prepare but not autonomously approve account closures, and descriptive rules do not replace programmatic validation or IAM, 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 call a protected internal pricing API under the application's own credentials and to enable a customer service agent to prepare but not autonomously approve account closures, 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 131
Engineers are splitting an incident response service into two independently tested functions. Function A must meet a requirement that proprietary repository contents remain within an approved execution environment, while function B must reject a shell command that targets protected deployment secrets before execution. Which plan is sound?
- First component: Deploy agent workers close to controlled data sources. Second component: Choose isolated sandbox execution for untrusted code.
- First component: Use the Claude Agent SDK for an integrated tool-driven agent. Second component: Use PreToolUse gating for dangerous operations.
- First component: Choose isolated sandbox execution for untrusted code. Second component: Use the Claude Agent SDK for an integrated tool-driven agent.
- First component: Deploy agent workers close to controlled data sources. Second component: Use PreToolUse gating for dangerous operations.
Explanation
Correct Answer
D:The first proposal fits its assigned role because a self-hosted harness can enforce infrastructure, network, and retention controls for the surrounding tools, while the second proposal fits its assigned role because a pre-execution hook can examine a proposed action and block it through a deterministic policy; their respective acceptance criteria are to meet a requirement that proprietary repository contents remain within an approved execution environment and to reject a shell command that targets protected deployment secrets before execution.
Incorrect Answers
A:The first proposal fits its assigned role because a self-hosted harness can enforce infrastructure, network, and retention controls for the surrounding tools, while the second proposal serves the different purpose to let an agent run generated scripts without exposing production host files instead of the assigned second role because a restricted execution environment limits filesystem, network, and resource privileges; their respective acceptance criteria are to meet a requirement that proprietary repository contents remain within an approved execution environment and to reject a shell command that targets protected deployment secrets before execution.
B:The first proposal serves the different purpose to build a coding agent that needs managed tool execution, sessions, and event handling instead of the assigned first role because the SDK provides agent-loop primitives and integrations that remove much of the custom harness work, while the second proposal fits its assigned role because a pre-execution hook can examine a proposed action and block it through a deterministic policy; their respective acceptance criteria are to meet a requirement that proprietary repository contents remain within an approved execution environment and to reject a shell command that targets protected deployment secrets before execution.
C:The first proposal serves the different purpose to let an agent run generated scripts without exposing production host files instead of the assigned first role because a restricted execution environment limits filesystem, network, and resource privileges, while the second proposal serves the different purpose to build a coding agent that needs managed tool execution, sessions, and event handling instead of the assigned second role because the SDK provides agent-loop primitives and integrations that remove much of the custom harness work; their respective acceptance criteria are to meet a requirement that proprietary repository contents remain within an approved execution environment and to reject a shell command that targets protected deployment secrets before execution.
Question 132
An output validation review requires two independent, explicitly defined outcomes: the first workstream must fit a large knowledge base into the context of a customer answer; a second must stop stale log dumps from crowding out new evidence during an agent session. Which TWO actions directly provide these capabilities? Choose TWO.
- Anchor summaries to source identifiers
- Refresh stale evidence before reuse
- Prioritize high-signal retrieved passages
- Prune obsolete tool outputs
- Reserve space for the model's output
Explanation
Correct Answers
C:Ranking by relevance and authority places the most useful evidence ahead of incidental matches; this directly supports task 1 to fit a large knowledge base into the context of a customer answer, not the separate task or tasks to stop stale log dumps from crowding out new evidence during an agent session; dumping every document into the prompt can obscure the actual answer.
D:Tool-result pruning removes redundant bulky observations after their findings are retained; this directly supports task 2 to stop stale log dumps from crowding out new evidence during an agent session, not the separate task or tasks to fit a large knowledge base into the context of a customer answer; keeping all raw logs forever does not guarantee better reasoning.
Incorrect Answers
A:Stable document IDs and passage references preserve provenance beyond the shortened working memory; this implements the different outcome to enable a reviewer to reconstruct evidence after context has been compacted rather than the required tasks to fit a large knowledge base into the context of a customer answer and to stop stale log dumps from crowding out new evidence during an agent session, and a bare compressed claim cannot show which source established it, so it is not a direct solution to any of the assigned workstreams.
B:Freshness checks determine whether a saved passage remains authoritative for a dynamic fact; this implements the different outcome to avoid answering a pricing question from last month's retrieved cache rather than the required tasks to fit a large knowledge base into the context of a customer answer and to stop stale log dumps from crowding out new evidence during an agent session, and high semantic similarity says nothing about timeliness, so it is not a direct solution to any of the assigned workstreams.
E:Input assembly must leave room for the requested completion and any needed reasoning; this implements the different outcome to avoid using the entire context budget on retrieved input before generating a response rather than the required tasks to fit a large knowledge base into the context of a customer answer and to stop stale log dumps from crowding out new evidence during an agent session, and a technically accepted long prompt can still prevent a useful-length answer, so it is not a direct solution to any of the assigned workstreams.
Question 133
Two separate tickets in a data pipeline backlog require different controls. One needs to keep follow-up requests linked without exposing another tenant's messages; the other needs to avoid exposing raw production database credentials directly to an agent. Which implementation pair is correct?
- First component: Return user-friendly error states. Second component: Build a typed output envelope.
- First component: Build a typed output envelope. Second component: Isolate tools behind a service layer.
- First component: Create one session per user conversation. Second component: Return user-friendly error states.
- First component: Create one session per user conversation. Second component: Isolate tools behind a service layer.
Explanation
Correct Answer
D:The first proposal fits its assigned role because session scoping ensures history and tool results belong to the correct principal, while the second proposal fits its assigned role because a mediation service validates identity, arguments, authorization, and audit logs for each requested action; their respective acceptance criteria are to keep follow-up requests linked without exposing another tenant's messages and to avoid exposing raw production database credentials directly to an agent.
Incorrect Answers
A:The first proposal serves the different purpose to explain service limits or tool outages without leaking stack traces or secrets instead of the assigned first role because structured error mapping presents actionable feedback while retaining detailed diagnostics in protected logs, while the second proposal serves the different purpose to send extracted contract fields into a database that rejects unexpected shapes instead of the assigned second role because an explicit schema with required fields, types, and allowed values makes integration errors detectable; their respective acceptance criteria are to keep follow-up requests linked without exposing another tenant's messages and to avoid exposing raw production database credentials directly to an agent.
B:The first proposal serves the different purpose to send extracted contract fields into a database that rejects unexpected shapes instead of the assigned first role because an explicit schema with required fields, types, and allowed values makes integration errors detectable, while the second proposal fits its assigned role because a mediation service validates identity, arguments, authorization, and audit logs for each requested action; their respective acceptance criteria are to keep follow-up requests linked without exposing another tenant's messages and to avoid exposing raw production database credentials directly to an agent.
C:The first proposal fits its assigned role because session scoping ensures history and tool results belong to the correct principal, while the second proposal serves the different purpose to explain service limits or tool outages without leaking stack traces or secrets instead of the assigned second role because structured error mapping presents actionable feedback while retaining detailed diagnostics in protected logs; their respective acceptance criteria are to keep follow-up requests linked without exposing another tenant's messages and to avoid exposing raw production database credentials directly to an agent.
Question 134
A technical lead requests a solution that can catch schema and policy failures after a model or tool update within a Claude application project. Which design should be chosen?
- Instrument production observability
- Implement automated regression tests
- Assign an owner to every integration
- Separate development, staging, and production environments
Explanation
Correct Answer
B:Repeatable cases compare changes against documented acceptance criteria; this behavior directly supports the need to catch schema and policy failures after a model or tool update. A single successful demonstration provides little protection against regressions.
Incorrect Answers
A:Logs, metrics, traces, and alert thresholds reveal operational failures at the right layer; the supported outcome is to identify rising error rates and latency after launch, whereas this scenario requires the ability to catch schema and policy failures after a model or tool update; user feedback alone arrives too late for timely diagnosis.
C:Ownership enables rotation, monitoring, change approval, and retirement decisions; despite being useful when a service must prevent abandoned connectors from retaining access after an application is transferred, it is not a solution to catch schema and policy failures after a model or tool update; an undocumented connector can remain privileged beyond its operational need.
D:For the alternate objective to exercise an agent's database tool without touching live customer records, environment boundaries keep credentials, data, and writes isolated during testing; the feature here must catch schema and policy failures after a model or tool update, and reusing production tokens in a local experiment defeats the separation.
Question 135
A team maintaining an agent governance needs to prevent a compromised tool step from sending internal files to arbitrary destinations. Which implementation most directly satisfies the requirement?
- Limit external network egress for agents
- Treat tool and retrieval text as untrusted
- Use output escaping to prevent XSS
- Apply retention limits to diagnostic logs
- Validate tool arguments server-side
Explanation
Correct Answer
A:Network policy restricts which services a tool process can contact; this behavior directly supports the need to prevent a compromised tool step from sending internal files to arbitrary destinations. A natural-language prohibition cannot enforce packet-level isolation.
Incorrect Answers
B:For the alternate objective to stop an injected wiki page from asking an assistant to export credentials, lower-trust content is evidence only; control decisions remain with authenticated policy and tool gates; the feature here must prevent a compromised tool step from sending internal files to arbitrary destinations, and a model can be persuaded by source text unless boundaries are enforced.
C:Context-aware encoding prevents text from being interpreted as active HTML or JavaScript; this applies to the different goal to display generated customer notes safely in a browser, not the requirement to prevent a compromised tool step from sending internal files to arbitrary destinations; an instruction to the model not to write scripts is not a rendering control.
D:When an application needs to reduce exposure of customer documents copied into application telemetry, data minimization and controlled retention reduce the duration and audience of sensitive log content; the current capability must instead prevent a compromised tool step from sending internal files to arbitrary destinations, while an internal log is still a data store requiring access controls.
E:Backend authorization checks must bind the requested operation to the caller's entitlements; the supported outcome is to block a generated database query that targets another tenant's records, whereas this scenario requires the ability to prevent a compromised tool step from sending internal files to arbitrary destinations; well-formed arguments are not proof of permission.
Question 136
A production API solution has parallel requirements with separate owners. Workstream one must maintain multiple Claude entry points without drifting retry and logging behavior; workstream two must test application logic without issuing real model calls or payments. Which combination meets both?
- First component: Version APIs and response schemas. Second component: Apply dependency injection for external clients.
- First component: Refactor duplicated request orchestration. Second component: Correlate distributed traces across services.
- First component: Refactor duplicated request orchestration. Second component: Apply dependency injection for external clients.
- First component: Correlate distributed traces across services. Second component: Version APIs and response schemas.
Explanation
Correct Answer
C:The first proposal fits its assigned role because a shared, tested client layer centralizes cross-cutting concerns and reduces inconsistent fixes, while the second proposal fits its assigned role because injected interfaces allow deterministic stubs and reduce coupling to vendors; their respective acceptance criteria are to maintain multiple Claude entry points without drifting retry and logging behavior and to test application logic without issuing real model calls or payments.
Incorrect Answers
A:The first proposal serves the different purpose to allow older application clients to keep working while a new extraction field is introduced instead of the assigned first role because compatibility-aware contracts and versioned behavior protect consumers from surprise breaking changes, while the second proposal fits its assigned role because injected interfaces allow deterministic stubs and reduce coupling to vendors; their respective acceptance criteria are to maintain multiple Claude entry points without drifting retry and logging behavior and to test application logic without issuing real model calls or payments.
B:The first proposal fits its assigned role because a shared, tested client layer centralizes cross-cutting concerns and reduces inconsistent fixes, while the second proposal serves the different purpose to debug where a request stalls between the web UI, queue, and Claude gateway instead of the assigned second role because a trace identifier links timing and errors through multiple process boundaries; their respective acceptance criteria are to maintain multiple Claude entry points without drifting retry and logging behavior and to test application logic without issuing real model calls or payments.
D:The first proposal serves the different purpose to debug where a request stalls between the web UI, queue, and Claude gateway instead of the assigned first role because a trace identifier links timing and errors through multiple process boundaries, while the second proposal serves the different purpose to allow older application clients to keep working while a new extraction field is introduced instead of the assigned second role because compatibility-aware contracts and versioned behavior protect consumers from surprise breaking changes; their respective acceptance criteria are to maintain multiple Claude entry points without drifting retry and logging behavior and to test application logic without issuing real model calls or payments.
Question 137
A design meeting for a workflow automation platform concludes that two isolated services require different behavior: one must avoid charging a customer twice after a worker retries a timed-out payment action, and another must meet a requirement that proprietary repository contents remain within an approved execution environment. Which arrangement meets the specification?
- First component: Use a durable session store. Second component: Deploy agent workers close to controlled data sources.
- First component: Add idempotency keys to side-effecting operations. Second component: Choose isolated sandbox execution for untrusted code.
- First component: Add idempotency keys to side-effecting operations. Second component: Deploy agent workers close to controlled data sources.
- First component: Choose isolated sandbox execution for untrusted code. Second component: Use a durable session store.
Explanation
Correct Answer
C:The first proposal fits its assigned role because application-generated operation identifiers allow the downstream service to recognize a repeated request, while the second proposal fits its assigned role because a self-hosted harness can enforce infrastructure, network, and retention controls for the surrounding tools; their respective acceptance criteria are to avoid charging a customer twice after a worker retries a timed-out payment action and to meet a requirement that proprietary repository contents remain within an approved execution environment.
Incorrect Answers
A:The first proposal serves the different purpose to recover agent state when a container restarts during a lengthy customer case instead of the assigned first role because external persistence holds the conversation and workflow metadata needed to resume safely, while the second proposal fits its assigned role because a self-hosted harness can enforce infrastructure, network, and retention controls for the surrounding tools; their respective acceptance criteria are to avoid charging a customer twice after a worker retries a timed-out payment action and to meet a requirement that proprietary repository contents remain within an approved execution environment.
B:The first proposal fits its assigned role because application-generated operation identifiers allow the downstream service to recognize a repeated request, while the second proposal serves the different purpose to let an agent run generated scripts without exposing production host files instead of the assigned second role because a restricted execution environment limits filesystem, network, and resource privileges; their respective acceptance criteria are to avoid charging a customer twice after a worker retries a timed-out payment action and to meet a requirement that proprietary repository contents remain within an approved execution environment.
D:The first proposal serves the different purpose to let an agent run generated scripts without exposing production host files instead of the assigned first role because a restricted execution environment limits filesystem, network, and resource privileges, while the second proposal serves the different purpose to recover agent state when a container restarts during a lengthy customer case instead of the assigned second role because external persistence holds the conversation and workflow metadata needed to resume safely; their respective acceptance criteria are to avoid charging a customer twice after a worker retries a timed-out payment action and to meet a requirement that proprietary repository contents remain within an approved execution environment.
Question 138
A team maintaining a privacy controls needs to detect unexpected API key activity from an unfamiliar process. Which implementation most directly satisfies the requirement?
- Audit sensitive credential use
- Authorize by requested resource and action
- Rotate a compromised API key promptly
- Use short-lived scoped credentials where supported
Explanation
Correct Answer
A:Usage and access logs support incident investigation and anomaly response; this behavior directly supports the need to detect unexpected API key activity from an unfamiliar process. A successful API call alone does not establish that use was authorized.
Incorrect Answers
B:Authorization checks must evaluate scope for each operation, not only whether a login exists; this applies to the different goal to deny access to a different customer's records even when the caller is authenticated, not the requirement to detect unexpected API key activity from an unfamiliar process; successful authentication is not equivalent to unrestricted data access.
C:When an application needs to cut off an exposed token discovered in CI logs, revocation plus replacement ends use of the leaked credential and requires checking for misuse; the current capability must instead detect unexpected API key activity from an unfamiliar process, while merely deleting a log message leaves the old token valid.
D:Expiring identity grants reduce the time and permissions available to an attacker; the supported outcome is to limit the impact of a temporary worker credential being stolen, whereas this scenario requires the ability to detect unexpected API key activity from an unfamiliar process; permanent shared administrator keys expand blast radius.
Question 139
During acceptance testing for a Claude application service, one requirement is to prevent abandoned connectors from retaining access after an application is transferred. What should the developer implement?
- Separate development, staging, and production environments
- Assign an owner to every integration
- Implement automated regression tests
- Define a staged release with rollback
Explanation
Correct Answer
B:Ownership enables rotation, monitoring, change approval, and retirement decisions; this behavior directly supports the need to prevent abandoned connectors from retaining access after an application is transferred. An undocumented connector can remain privileged beyond its operational need.
Incorrect Answers
A:When an application needs to exercise an agent's database tool without touching live customer records, environment boundaries keep credentials, data, and writes isolated during testing; the current capability must instead prevent abandoned connectors from retaining access after an application is transferred, while reusing production tokens in a local experiment defeats the separation.
C:This approach serves the separate need to catch schema and policy failures after a model or tool update because repeatable cases compare changes against documented acceptance criteria; it does not establish the outcome to prevent abandoned connectors from retaining access after an application is transferred, and a single successful demonstration provides little protection against regressions.
D:Gradual rollout and versioned rollback allow metrics to be compared while limiting impact; despite being useful when a service must deploy a prompt change without exposing every customer to an untested regression, it is not a solution to prevent abandoned connectors from retaining access after an application is transferred; editing the only production prompt in place makes reversions harder.
Question 140
Engineers are splitting a latency optimization service into two independently tested functions. Function A must spend high-capability tokens only when validation indicates complexity, while function B must avoid underbudgeting a report-writing feature whose completions are much longer than inputs. Which plan is sound?
- First component: Measure cache read tokens. Second component: Estimate output length before scaling.
- First component: Use model routing for simple versus hard cases. Second component: Set maximum step counts for agents.
- First component: Set maximum step counts for agents. Second component: Measure cache read tokens.
- First component: Use model routing for simple versus hard cases. Second component: Estimate output length before scaling.
Explanation
Correct Answer
D:The first proposal fits its assigned role because conditional routing improves blended cost without discarding an escalation path, while the second proposal fits its assigned role because cost models should include the generated token volume and task-specific variance; their respective acceptance criteria are to spend high-capability tokens only when validation indicates complexity and to avoid underbudgeting a report-writing feature whose completions are much longer than inputs.
Incorrect Answers
A:The first proposal serves the different purpose to verify that an optimization actually reuses an earlier prompt prefix instead of the assigned first role because usage fields show how much eligible cached input was read rather than recomputed, while the second proposal fits its assigned role because cost models should include the generated token volume and task-specific variance; their respective acceptance criteria are to spend high-capability tokens only when validation indicates complexity and to avoid underbudgeting a report-writing feature whose completions are much longer than inputs.
B:The first proposal fits its assigned role because conditional routing improves blended cost without discarding an escalation path, while the second proposal serves the different purpose to contain cumulative token and tool cost across a recursive research session instead of the assigned second role because a bound on iterations prevents repeated generation and calls from growing without limit; their respective acceptance criteria are to spend high-capability tokens only when validation indicates complexity and to avoid underbudgeting a report-writing feature whose completions are much longer than inputs.
C:The first proposal serves the different purpose to contain cumulative token and tool cost across a recursive research session instead of the assigned first role because a bound on iterations prevents repeated generation and calls from growing without limit, while the second proposal serves the different purpose to verify that an optimization actually reuses an earlier prompt prefix instead of the assigned second role because usage fields show how much eligible cached input was read rather than recomputed; their respective acceptance criteria are to spend high-capability tokens only when validation indicates complexity and to avoid underbudgeting a report-writing feature whose completions are much longer than inputs.