Question 420
For a Claude inference cost review, a design is acceptable only if it can identify why verbose generated reports cost more despite unchanged inputs while also being able to choose a cheaper model only when the workload-level economics improve. Which paired implementation is sufficient?
- Perform both steps: (1) Design for stochastic output and evaluate behavior over representative repeats; (2) compare quality-adjusted cost per accepted result across candidate models.
- Perform both steps: (1) Design for stochastic output and evaluate behavior over representative repeats; (2) alert on unusual per-job token and tool-call growth.
- Perform both steps: (1) Track output tokens separately from input tokens; (2) compare quality-adjusted cost per accepted result across candidate models.
- Perform both steps: (1) Track output tokens separately from input tokens; (2) alert on unusual per-job token and tool-call growth.
Explanation
Correct Answer
C:The first change enables the team to identify why verbose generated reports cost more despite unchanged inputs because generation tokens are metered independently and can have different pricing, so response verbosity affects unit economics; the second addresses the separate need to choose a cheaper model only when the workload-level economics improve because a lower token price can be offset by retries, higher failure rates, or excessive output, so accepted-result cost is the useful measure.
Incorrect Answers
A:The second step meets the requirement to choose a cheaper model only when the workload-level economics improve because a lower token price can be offset by retries, higher failure rates, or excessive output, so accepted-result cost is the useful measure; the first step would only help to account for different plausible answers from the same input on repeated runs because sampling and model execution can produce variation; important correctness must be checked rather than assumed identical on every call, leaving the separate requirement to identify why verbose generated reports cost more despite unchanged inputs unmet.
B:Both steps target different results—account for different plausible answers from the same input on repeated runs and detect a looping agent that suddenly multiplies token consumption—because sampling and model execution can produce variation; important correctness must be checked rather than assumed identical on every call and monitoring consumption across an entire agent run catches runaway workflows that per-request logs may not reveal; the required capabilities are instead to identify why verbose generated reports cost more despite unchanged inputs and to choose a cheaper model only when the workload-level economics improve.
D:The first step is appropriate for the requirement to identify why verbose generated reports cost more despite unchanged inputs because generation tokens are metered independently and can have different pricing, so response verbosity affects unit economics; the other step instead aims to detect a looping agent that suddenly multiplies token consumption because monitoring consumption across an entire agent run catches runaway workflows that per-request logs may not reveal, which does not meet the second requirement to choose a cheaper model only when the workload-level economics improve.
Question 421
A design review for a privileged agent security review asks for support to detect unauthorized modifications to a tool result before it reaches the model, together with a way to protect customer identity fields when only document topics are needed for classification. Select the two appropriate changes. Choose TWO.
- Verify origin and integrity of sensitive external data where supported.
- Pause execution for explicit approval at the high-impact action boundary.
- Remove or mask unnecessary personal information before model submission.
- Check resource-level authorization for every sensitive tool request.
- Enforce outbound destination allowlists at the tool execution layer.
Explanation
Correct Answers
A:To detect unauthorized modifications to a tool result before it reaches the model as part of a design that must also protect customer identity fields when only document topics are needed for classification, trusted decisions require confidence that tool outputs were not altered; transport and signature checks complement content validation; this is why the action is one of the two required selections.
C:To protect customer identity fields when only document topics are needed for classification as part of a design that must also detect unauthorized modifications to a tool result before it reaches the model, minimization limits exposure while preserving task inputs; relying on the model to ignore visible PII does not provide the same protection; this is why the action is one of the two required selections.
Incorrect Answers
B:This alternative serves the separate goal to require a person to approve an irreversible administrative change because human approval is a policy-enforced transition, not a note written in the model's plan or a message displayed afterward; the required outcomes are instead to detect unauthorized modifications to a tool result before it reaches the model and to protect customer identity fields when only document topics are needed for classification.
D:This alternative serves the separate goal to stop an authenticated user from reading another tenant's records via a tool because proving identity does not grant access to every object; the backend must enforce tenant and resource permissions separately; the required outcomes are instead to detect unauthorized modifications to a tool result before it reaches the model and to protect customer identity fields when only document topics are needed for classification.
E:This alternative serves the separate goal to stop a model-requested webhook from sending sensitive case text to an unapproved host because network restrictions prevent data egress even if injected instructions convince a model to request the call; the required outcomes are instead to detect unauthorized modifications to a tool result before it reaches the model and to protect customer identity fields when only document topics are needed for classification.
Question 422
To ship an MCP integration server, the application must help a client learn which callable and informational capabilities a server offers and make reference material available for clients to read as context. Which pair of listed actions is necessary for those stated capabilities? Choose TWO.
- Write a custom tool whose handler runs in the application.
- Implement the appropriate MCP capability listing and metadata endpoints.
- Create a Skill containing the procedure and supporting instructions.
- Expose the manual as an MCP resource rather than a tool.
- Expose the material as an MCP resource with a stable identifier.
Explanation
Correct Answers
B:To help a client learn which callable and informational capabilities a server offers as part of a design that must also make reference material available for clients to read as context, discovery advertises tools, resources, or prompts so clients can use supported contracts rather than guessing names or arguments; this is why the action is one of the two required selections.
E:To make reference material available for clients to read as context as part of a design that must also help a client learn which callable and informational capabilities a server offers, resources represent consumable data, contrasting with tools that invoke an operation or prompts that package reusable instructions; this is why the action is one of the two required selections.
Incorrect Answers
A:This alternative serves the separate goal to invoke a proprietary internal pricing database unavailable through a built-in feature because custom tools expose application-specific actions while retaining control over authentication, validation, and network access; the required outcomes are instead to help a client learn which callable and informational capabilities a server offers and to make reference material available for clients to read as context.
C:This alternative serves the separate goal to make a repeatable multi-step code-review methodology available to an agent because a Skill teaches how to perform a task, whereas a tool is the executable capability that performs an external operation; the required outcomes are instead to help a client learn which callable and informational capabilities a server offers and to make reference material available for clients to read as context.
D:This alternative serves the separate goal to supply a read-only policy manual to an MCP client without giving it an action endpoint because a resource is contextual data; representing a passive document as an imperative function adds unnecessary execution semantics; the required outcomes are instead to help a client learn which callable and informational capabilities a server offers and to make reference material available for clients to read as context.
Question 423
In a customer-facing AI feature, testing reveals that the system cannot identify which search services and internal systems are critical to the workflow and cannot translate a request for citation-backed answers into observable application behavior. Select the two targeted actions that close these gaps. Choose TWO.
- Document permitted processing regions and data-handling obligations before selecting a deployment.
- Define measurable answer-format and citation acceptance criteria.
- List upstream systems with permissions, quotas, and failure behavior.
- Return tool_result blocks containing the corresponding tool_use_id.
- Specify expected concurrency and enforce admission or queueing limits.
Explanation
Correct Answers
B:To translate a request for citation-backed answers into observable application behavior as part of a design that must also identify which search services and internal systems are critical to the workflow, a functional requirement specifies what the user must receive and how success is evaluated, not merely that the team will use a language model; this is why the action is one of the two required selections.
C:To identify which search services and internal systems are critical to the workflow as part of a design that must also translate a request for citation-backed answers into observable application behavior, a dependency inventory exposes the operational and authorization constraints that a purely model-centered design misses; this is why the action is one of the two required selections.
Incorrect Answers
A:This alternative serves the separate goal to keep sensitive case material within approved processing and retention boundaries because legal or organizational constraints must inform provider routing, logging, and retention rather than being added after data already flows; the required outcomes are instead to identify which search services and internal systems are critical to the workflow and to translate a request for citation-backed answers into observable application behavior.
D:This alternative serves the separate goal to connect execution output to the exact tool request Claude made because tool correlation preserves call-to-result semantics in the message sequence, especially when several tools were requested together; the required outcomes are instead to identify which search services and internal systems are critical to the workflow and to translate a request for citation-backed answers into observable application behavior.
E:This alternative serves the separate goal to design for a documented burst of simultaneous requests without uncontrolled overload because capacity planning separates average traffic from peak demand and prevents upstream rate limits from becoming cascading application failures; the required outcomes are instead to identify which search services and internal systems are critical to the workflow and to translate a request for citation-backed answers into observable application behavior.
Question 424
For a model-family migration plan, the missing capability is to cut cost for a nightly noninteractive summarization workload. What should the developers implement?
- Evaluate a Sonnet-family model against a realistic task set.
- Route simple requests to a cheaper evaluated tier and escalate difficult ones.
- Evaluate Message Batches instead of synchronous requests.
- Measure or count input and output tokens with the supported tokenizer or API.
Explanation
Correct Answer
C:To cut cost for a nightly noninteractive summarization workload in this model-family migration plan, asynchronous batches can offer discounted processing when the application can tolerate deferred completion and result collection; the proposed implementation directly supplies that capability.
Incorrect Answers
A:The proposed alternative helps to deliver moderately complex development assistance with balanced quality, cost, and latency because a balanced model tier often offers useful capability without the latency or cost of the most expensive tier; evaluation should drive final selection; that mechanism does not meet the different requirement to cut cost for a nightly noninteractive summarization workload.
B:The proposed alternative helps to minimize average model cost when most requests are easy but some require deeper reasoning because cascading matches compute to task difficulty but requires a reliable escalation rule and measurement of misrouting; that mechanism does not meet the different requirement to cut cost for a nightly noninteractive summarization workload.
D:The proposed alternative helps to estimate model usage from encoded text rather than counting words because billing and context consumption track tokens, which do not map one-to-one to English words or characters; that mechanism does not meet the different requirement to cut cost for a nightly noninteractive summarization workload.
Question 425
Two acceptance tests block a document investigation agent: one requires the ability to require precise control over message assembly, tool dispatch, and stopping behavior; the other requires the ability to continue a paused multi-step task with its legitimate prior context. Which plan clears both tests?
- Perform both steps: (1) Consume the agent runtime's streamed event sequence; (2) persist and resume the agent session using supported state handling.
- Perform both steps: (1) Consume the agent runtime's streamed event sequence; (2) give an agent a bounded objective and a tool-selection loop.
- Perform both steps: (1) Implement an application-owned Messages API agent loop; (2) persist and resume the agent session using supported state handling.
- Perform both steps: (1) Implement an application-owned Messages API agent loop; (2) give an agent a bounded objective and a tool-selection loop.
Explanation
Correct Answer
C:The first change enables the team to require precise control over message assembly, tool dispatch, and stopping behavior because a custom loop explicitly owns each model call, tool result, state update, and budget decision, trading flexibility for more engineering responsibility; the second addresses the separate need to continue a paused multi-step task with its legitimate prior context because session continuity preserves relevant work across invocations; discarding session state can cause repeated tools or lost intermediate decisions.
Incorrect Answers
A:The second step meets the requirement to continue a paused multi-step task with its legitimate prior context because session continuity preserves relevant work across invocations; discarding session state can cause repeated tools or lost intermediate decisions; the first step would only help to deliver intermediate agent events to the client as they occur because event-driven consumption lets the application display progress and inspect tool actions without waiting for the entire multi-step run to finish, leaving the separate requirement to require precise control over message assembly, tool dispatch, and stopping behavior unmet.
B:Both steps target different results—deliver intermediate agent events to the client as they occur and pursue an open-ended investigation where the next action depends on intermediate evidence—because event-driven consumption lets the application display progress and inspect tool actions without waiting for the entire multi-step run to finish and agentic control lets the model choose among permitted steps as observations arrive; budgets and termination conditions still bound its autonomy; the required capabilities are instead to require precise control over message assembly, tool dispatch, and stopping behavior and to continue a paused multi-step task with its legitimate prior context.
D:The first step is appropriate for the requirement to require precise control over message assembly, tool dispatch, and stopping behavior because a custom loop explicitly owns each model call, tool result, state update, and budget decision, trading flexibility for more engineering responsibility; the other step instead aims to pursue an open-ended investigation where the next action depends on intermediate evidence because agentic control lets the model choose among permitted steps as observations arrive; budgets and termination conditions still bound its autonomy, which does not meet the second requirement to continue a paused multi-step task with its legitimate prior context.
Question 426
A change request for a high-volume Messages API gateway combines two needs: recognize that the model is asking the application to execute a client tool, and ask questions about information in a user-provided PDF through a supported document input. Which implementation pair addresses both, not just one?
- Perform both steps: (1) Branch on stop_reason tool_use and dispatch the declared calls; (2) attach the pdf using the supported document content form.
- Perform both steps: (1) Branch on stop_reason tool_use and dispatch the declared calls; (2) cache an eligible unchanged prefix with cache_control.
- Perform both steps: (1) Assign and preserve unique custom identifiers for batched requests; (2) cache an eligible unchanged prefix with cache_control.
- Perform both steps: (1) Assign and preserve unique custom identifiers for batched requests; (2) attach the pdf using the supported document content form.
Explanation
Correct Answer
A:The first change enables the team to recognize that the model is asking the application to execute a client tool because a tool-use stop requires host-side handling and a follow-up message containing results, not presenting the call as final prose; the second addresses the separate need to ask questions about information in a user-provided PDF through a supported document input because document analysis depends on actual supported content blocks and size limits; a local file path alone is not accessible to the remote model.
Incorrect Answers
B:The first step is appropriate for the requirement to recognize that the model is asking the application to execute a client tool because a tool-use stop requires host-side handling and a follow-up message containing results, not presenting the call as final prose; the other step instead aims to reuse a long stable instruction or document prefix over repeated requests because prompt caching reduces repeated input processing when the prefix matches; changes before the cache boundary can prevent a hit, which does not meet the second requirement to ask questions about information in a user-provided PDF through a supported document input.
C:Both steps target different results—map out-of-order batch outputs back to source records reliably and reuse a long stable instruction or document prefix over repeated requests—because asynchronous results need stable correlation keys; relying on processing order can attach an output to the wrong input and prompt caching reduces repeated input processing when the prefix matches; changes before the cache boundary can prevent a hit; the required capabilities are instead to recognize that the model is asking the application to execute a client tool and to ask questions about information in a user-provided PDF through a supported document input.
D:The second step meets the requirement to ask questions about information in a user-provided PDF through a supported document input because document analysis depends on actual supported content blocks and size limits; a local file path alone is not accessible to the remote model; the first step would only help to map out-of-order batch outputs back to source records reliably because asynchronous results need stable correlation keys; relying on processing order can attach an output to the wrong input, leaving the separate requirement to recognize that the model is asking the application to execute a client tool unmet.
Question 427
The owner of a privileged agent security review needs to stop an authenticated user from reading another tenant's records via a tool as well as resist instructions embedded in a retrieved support ticket that demand secret disclosure. Which two changes belong in the minimum remediation plan? Choose TWO.
- Treat retrieved content as untrusted and prevent its instructions from controlling tools.
- Check resource-level authorization for every sensitive tool request.
- Record result metadata in a post-tool hook.
- Retrieve keys through a managed secrets facility at runtime.
- Use a server-side workload identity or scoped service credential.
Explanation
Correct Answers
A:To resist instructions embedded in a retrieved support ticket that demand secret disclosure as part of a design that must also stop an authenticated user from reading another tenant's records via a tool, source material is data rather than authority; enforcing tool permissions and separating roles limits an attacker's ability to redirect execution; this is why the action is one of the two required selections.
B:To stop an authenticated user from reading another tenant's records via a tool as part of a design that must also resist instructions embedded in a retrieved support ticket that demand secret disclosure, proving identity does not grant access to every object; the backend must enforce tenant and resource permissions separately; this is why the action is one of the two required selections.
Incorrect Answers
C:This alternative serves the separate goal to capture outcomes of agent command execution for security review because post-execution hooks can log and inspect what happened; they are unsuitable as the sole barrier to a side effect that already occurred; the required outcomes are instead to stop an authenticated user from reading another tenant's records via a tool and to resist instructions embedded in a retrieved support ticket that demand secret disclosure.
D:This alternative serves the separate goal to keep Claude API credentials out of repository source and deployment images because central secret handling improves rotation and auditing and avoids accidental credential disclosure through source control; the required outcomes are instead to stop an authenticated user from reading another tenant's records via a tool and to resist instructions embedded in a retrieved support ticket that demand secret disclosure.
E:This alternative serves the separate goal to stop frontend users from receiving unrestricted backend tool credentials because service identities tie privileges to a runtime workload and avoid exposing long-lived secrets to browsers or untrusted users; the required outcomes are instead to stop an authenticated user from reading another tenant's records via a tool and to resist instructions embedded in a retrieved support ticket that demand secret disclosure.
Question 428
In a token-consumption optimization sprint, testing reveals that the system cannot build bidirectional persistent event communication where the chosen platform supports it and cannot detect changed library behavior before rolling a new version into production. Select the two targeted actions that close these gaps. Choose TWO.
- Check each response block's type before extracting its fields.
- Use a WebSocket protocol only when the target endpoint actually offers one.
- Alert on unusual per-job token and tool-call growth.
- Branch on HTTP status and decode the provider's error structure.
- Run contract and integration tests against the upgraded SDK in staging.
Explanation
Correct Answers
B:To build bidirectional persistent event communication where the chosen platform supports it as part of a design that must also detect changed library behavior before rolling a new version into production, webSockets can carry two-way events over a long-lived connection; they are not a drop-in replacement for every HTTP or SSE endpoint; this is why the action is one of the two required selections.
E:To detect changed library behavior before rolling a new version into production as part of a design that must also build bidirectional persistent event communication where the chosen platform supports it, client-library version shifts can affect serialization or errors even without changes to business prompts, so they deserve release validation; this is why the action is one of the two required selections.
Incorrect Answers
A:This alternative serves the separate goal to prevent a tool_use block from being displayed as if it were natural-language text because typed response objects can contain several block families; assuming every block has a text field causes logic errors; the required outcomes are instead to build bidirectional persistent event communication where the chosen platform supports it and to detect changed library behavior before rolling a new version into production.
C:This alternative serves the separate goal to detect a looping agent that suddenly multiplies token consumption because monitoring consumption across an entire agent run catches runaway workflows that per-request logs may not reveal; the required outcomes are instead to build bidirectional persistent event communication where the chosen platform supports it and to detect changed library behavior before rolling a new version into production.
D:This alternative serves the separate goal to stop interpreting non-2xx API responses as normal assistant messages because transport errors are not regular model completions, so application control flow must treat them separately; the required outcomes are instead to build bidirectional persistent event communication where the chosen platform supports it and to detect changed library behavior before rolling a new version into production.
Question 429
An operational review of a tool-calling Claude workflow reports distinct failures: it cannot prevent two similarly named lookup tools from being confused, and it cannot require or restrict tool use for a specific turn when the API supports that control. Which two repairs are required? Choose TWO.
- Configure a supported remote HTTP transport with authentication and deployment controls.
- Describe each tool's purpose, required inputs, and limitations clearly.
- Configure the supported tool_choice behavior for the request.
- Give each tool a timeout and controlled retry or failure result.
- Put an approval gate in the host-side write handler.
Explanation
Correct Answers
B:To prevent two similarly named lookup tools from being confused as part of a design that must also require or restrict tool use for a specific turn when the API supports that control, tool descriptions influence the model's selection, so overlapping or vague names make dispatch less reliable; this is why the action is one of the two required selections.
C:To require or restrict tool use for a specific turn when the API supports that control as part of a design that must also prevent two similarly named lookup tools from being confused, tool-selection controls influence whether Claude may call tools, but cannot grant permissions to the underlying systems; this is why the action is one of the two required selections.
Incorrect Answers
A:This alternative serves the separate goal to expose an MCP server to clients across a network boundary because remote connectivity needs network routing, identity, and authorization; a local stdio subprocess is not a remotely reachable service; the required outcomes are instead to prevent two similarly named lookup tools from being confused and to require or restrict tool use for a specific turn when the API supports that control.
D:This alternative serves the separate goal to stop a slow external system from trapping an agent loop indefinitely because tool execution is a separate latency source; bounded calls prevent one dependency from exhausting the entire agent budget; the required outcomes are instead to prevent two similarly named lookup tools from being confused and to require or restrict tool use for a specific turn when the API supports that control.
E:This alternative serves the separate goal to ensure an agent obtains explicit authorization before updating a customer record because an irreversible or sensitive write must be permitted by trusted application logic, not merely because the model selected a tool; the required outcomes are instead to prevent two similarly named lookup tools from being confused and to require or restrict tool use for a specific turn when the API supports that control.
Question 430
For a Python Claude API backend, the missing capability is to ensure refunds above a policy limit receive human sign-off. What should the developers implement?
- Apply content checks at suitable input and output boundaries.
- Use strict tool input validation for the exposed tool interface.
- Persist explicit job states with allowed transitions.
- Add application-owned approval logic around the refund tool.
Explanation
Correct Answer
D:To ensure refunds above a policy limit receive human sign-off in this Python Claude API backend, decisions that create financial side effects require deterministic authorization checks beyond a model's recommendation; the proposed implementation directly supplies that capability.
Incorrect Answers
A:The proposed alternative helps to filter risky user-visible content according to application policy because moderation is a defense layer applied around generation, not a substitute for authenticated tool access or scoped retrieval; that mechanism does not meet the different requirement to ensure refunds above a policy limit receive human sign-off.
B:The proposed alternative helps to reject tool arguments that fail a declared shape before tool execution because a strict schema constrains the arguments a model may emit, but application authorization and business-rule checks remain necessary; that mechanism does not meet the different requirement to ensure refunds above a policy limit receive human sign-off.
C:The proposed alternative helps to distinguish queued, running, waiting-for-approval, failed, and completed agent jobs because a state machine prevents an incomplete tool operation from being shown as successful and makes retries and cancellation predictable; that mechanism does not meet the different requirement to ensure refunds above a policy limit receive human sign-off.