Anthropic CCDV-F (Claude Certified Developer - Foundations) Practice Test 36

Question 350

A team maintaining a Python Claude API backend has two different engineering requirements: (1) handle a response containing reasoning-related blocks without confusing them with ordinary final text, and (2) support Claude models accessed through different supported cloud offerings. Which TWO independent changes directly meet them? Choose TWO.

  1. Use the provider-specific endpoint, authentication, and request conventions.
  2. Send instructions using the request's top-level system field.
  3. Parse response content by block type and follow current thinking-field rules.
  4. Detect refusal and apply the product's allowed fallback or messaging policy.
  5. Treat stop_reason end_turn as a completed assistant turn.

Explanation

Correct Answers

A:

To support Claude models accessed through different supported cloud offerings as part of a design that must also handle a response containing reasoning-related blocks without confusing them with ordinary final text, Claude API, Bedrock, and Vertex-style integrations are not interchangeable URLs; host authentication and service wrappers differ; this is why the action is one of the two required selections.

C:

To handle a response containing reasoning-related blocks without confusing them with ordinary final text as part of a design that must also support Claude models accessed through different supported cloud offerings, structured content may contain distinct block types and signatures; a parser that assumes only plain text can discard important protocol data; this is why the action is one of the two required selections.

Incorrect Answers

B:

This alternative serves the separate goal to set consistent developer-controlled behavior separately from untrusted user text because the system field separates higher-priority application guidance from user-message content and should not be simulated by ordinary quoted text; the required outcomes are instead to handle a response containing reasoning-related blocks without confusing them with ordinary final text and to support Claude models accessed through different supported cloud offerings.

D:

This alternative serves the separate goal to handle a safety refusal as a distinct application outcome because a refusal is not the same as a network failure; blindly retrying identical input may not produce an appropriate user experience; the required outcomes are instead to handle a response containing reasoning-related blocks without confusing them with ordinary final text and to support Claude models accessed through different supported cloud offerings.

E:

This alternative serves the separate goal to detect that the assistant has naturally finished a response because the completion signal is explicit in the response metadata; it should not be inferred solely from a sentence ending in punctuation; the required outcomes are instead to handle a response containing reasoning-related blocks without confusing them with ordinary final text and to support Claude models accessed through different supported cloud offerings.

Question 351

A change request for a Claude API workload benchmark combines two needs: invite stylistic variety during noncritical brainstorming, and favor more consistent word choices in a constrained transformation task when supported. Which implementation pair addresses both, not just one?

  1. Perform both steps: (1) Use more exploratory sampling where supported and measure quality; (2) compare both candidates on identical representative eval cases.
  2. Perform both steps: (1) Use more exploratory sampling where supported and measure quality; (2) tune sampling for lower variance while retaining output validation.
  3. Perform both steps: (1) Configure supported adaptive thinking behavior rather than forcing a fixed pattern; (2) tune sampling for lower variance while retaining output validation.
  4. Perform both steps: (1) Configure supported adaptive thinking behavior rather than forcing a fixed pattern; (2) compare both candidates on identical representative eval cases.

Explanation

Correct Answer

B:

The first change enables the team to invite stylistic variety during noncritical brainstorming because higher sampling randomness broadens candidate phrasing but can reduce predictability, so it suits ideation better than strict extraction; the second addresses the separate need to favor more consistent word choices in a constrained transformation task when supported because temperature influences token sampling; it cannot by itself guarantee truth, policy compliance, or reproducible structured output.

Incorrect Answers

A:

The first step is appropriate for the requirement to invite stylistic variety during noncritical brainstorming because higher sampling randomness broadens candidate phrasing but can reduce predictability, so it suits ideation better than strict extraction; the other step instead aims to choose between two supported model releases using evidence rather than reputation because holding inputs and scoring constant reveals quality and latency changes that generic model descriptions cannot settle, which does not meet the second requirement to favor more consistent word choices in a constrained transformation task when supported.

C:

The second step meets the requirement to favor more consistent word choices in a constrained transformation task when supported because temperature influences token sampling; it cannot by itself guarantee truth, policy compliance, or reproducible structured output; the first step would only help to allow supported models to vary internal reasoning effort according to task difficulty because adaptive reasoning responds to problem complexity, but supported modes differ by model and release and should not be assumed universal, leaving the separate requirement to invite stylistic variety during noncritical brainstorming unmet.

D:

Both steps target different results—allow supported models to vary internal reasoning effort according to task difficulty and choose between two supported model releases using evidence rather than reputation—because adaptive reasoning responds to problem complexity, but supported modes differ by model and release and should not be assumed universal and holding inputs and scoring constant reveals quality and latency changes that generic model descriptions cannot settle; the required capabilities are instead to invite stylistic variety during noncritical brainstorming and to favor more consistent word choices in a constrained transformation task when supported.

Question 352

For a multi-stage support automation, a design is acceptable only if it can keep a large source-reading task from crowding the coordinator's working context while also being able to have independent specialized reviewers inspect the same proposal from different perspectives. Which paired implementation is sufficient?

  1. Perform both steps: (1) Inspect the tool_use block, invoke the authorized handler, and return tool_result; (2) retrieve only task-relevant memories into the active context.
  2. Perform both steps: (1) Inspect the tool_use block, invoke the authorized handler, and return tool_result; (2) fan out to role-specific reviewers and aggregate disagreements.
  3. Perform both steps: (1) Delegate source analysis to a subagent and return a concise result; (2) fan out to role-specific reviewers and aggregate disagreements.
  4. Perform both steps: (1) Delegate source analysis to a subagent and return a concise result; (2) retrieve only task-relevant memories into the active context.

Explanation

Correct Answer

C:

The first change enables the team to keep a large source-reading task from crowding the coordinator's working context because isolated workspaces let a specialist process extensive material while the parent receives only task-relevant evidence instead of every intermediate token; the second addresses the separate need to have independent specialized reviewers inspect the same proposal from different perspectives because separating reviews allows conflicting evidence to surface before the supervisor combines findings; a single sequential review may miss independent critiques.

Incorrect Answers

A:

Both steps target different results—execute a requested internal business function inside the application's trusted process and stop old agent histories from burying the information needed for the current step—because client tools are executed by the application, not by merely declaring a schema; every returned result must be matched to its tool-use identifier and selective retrieval improves signal-to-noise and keeps token use bounded while retaining access to older information when required; the required capabilities are instead to keep a large source-reading task from crowding the coordinator's working context and to have independent specialized reviewers inspect the same proposal from different perspectives.

B:

The second step meets the requirement to have independent specialized reviewers inspect the same proposal from different perspectives because separating reviews allows conflicting evidence to surface before the supervisor combines findings; a single sequential review may miss independent critiques; the first step would only help to execute a requested internal business function inside the application's trusted process because client tools are executed by the application, not by merely declaring a schema; every returned result must be matched to its tool-use identifier, leaving the separate requirement to keep a large source-reading task from crowding the coordinator's working context unmet.

D:

The first step is appropriate for the requirement to keep a large source-reading task from crowding the coordinator's working context because isolated workspaces let a specialist process extensive material while the parent receives only task-relevant evidence instead of every intermediate token; the other step instead aims to stop old agent histories from burying the information needed for the current step because selective retrieval improves signal-to-noise and keeps token use bounded while retaining access to older information when required, which does not meet the second requirement to have independent specialized reviewers inspect the same proposal from different perspectives.

Question 353

The technical owner of a business workflow that calls Claude has to enable an administrator to determine how a particular AI decision was produced. A separate requirement is to state how much interruption a customer-facing AI feature can tolerate. Which two-part remediation covers both?

  1. Perform both steps: (1) Record a blameless timeline, root causes, and preventive controls; (2) define availability objectives with failure and recovery criteria.
  2. Perform both steps: (1) Record a blameless timeline, root causes, and preventive controls; (2) set realistic connection and response timeouts with an explicit failure path.
  3. Perform both steps: (1) Require structured request IDs, tool traces, and version metadata with privacy controls; (2) define availability objectives with failure and recovery criteria.
  4. Perform both steps: (1) Require structured request IDs, tool traces, and version metadata with privacy controls; (2) set realistic connection and response timeouts with an explicit failure path.

Explanation

Correct Answer

C:

The first change enables the team to enable an administrator to determine how a particular AI decision was produced because traceability depends on consistent operational evidence, not just retaining the final generated answer or a screenshot; the second addresses the separate need to state how much interruption a customer-facing AI feature can tolerate because reliability requirements determine fallback, retries, health monitoring, and degraded behavior; they are distinct from the response's linguistic quality.

Incorrect Answers

A:

The second step meets the requirement to state how much interruption a customer-facing AI feature can tolerate because reliability requirements determine fallback, retries, health monitoring, and degraded behavior; they are distinct from the response's linguistic quality; the first step would only help to learn from a tool-side effect caused by an unexpected model action because post-incident review should identify inadequate approvals and monitoring and turn findings into testable changes, rather than blaming the language model, leaving the separate requirement to enable an administrator to determine how a particular AI decision was produced unmet.

B:

Both steps target different results—learn from a tool-side effect caused by an unexpected model action and keep a stalled upstream model call from occupying application resources indefinitely—because post-incident review should identify inadequate approvals and monitoring and turn findings into testable changes, rather than blaming the language model and timeout handling caps resource exposure, while logging distinguishes slow provider response from a model-level refusal; the required capabilities are instead to enable an administrator to determine how a particular AI decision was produced and to state how much interruption a customer-facing AI feature can tolerate.

D:

The first step is appropriate for the requirement to enable an administrator to determine how a particular AI decision was produced because traceability depends on consistent operational evidence, not just retaining the final generated answer or a screenshot; the other step instead aims to keep a stalled upstream model call from occupying application resources indefinitely because timeout handling caps resource exposure, while logging distinguishes slow provider response from a model-level refusal, which does not meet the second requirement to state how much interruption a customer-facing AI feature can tolerate.

Question 354

An end-to-end test for a Claude inference cost review checks whether it can stop interpreting non-2xx API responses as normal assistant messages and protect model API traffic against an impersonated network endpoint. Which plan should be approved to satisfy both checks?

  1. Perform both steps: (1) Use the model's documented thinking capability within the latency budget; (2) keep https certificate and hostname verification enabled.
  2. Perform both steps: (1) Branch on HTTP status and decode the provider's error structure; (2) keep https certificate and hostname verification enabled.
  3. Perform both steps: (1) Use the model's documented thinking capability within the latency budget; (2) parse structured payloads with explicit failure handling.
  4. Perform both steps: (1) Branch on HTTP status and decode the provider's error structure; (2) parse structured payloads with explicit failure handling.

Explanation

Correct Answer

B:

The first change enables the team 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 second addresses the separate need to protect model API traffic against an impersonated network endpoint because verified TLS authenticates the service and protects data in transit; disabling validation defeats a key transport security property.

Incorrect Answers

A:

The second step meets the requirement to protect model API traffic against an impersonated network endpoint because verified TLS authenticates the service and protects data in transit; disabling validation defeats a key transport security property; the first step would only help to allocate additional reasoning effort to a complex multistep decision where supported because reasoning settings can change token consumption and response latency and should be evaluated for the particular workload, leaving the separate requirement to stop interpreting non-2xx API responses as normal assistant messages unmet.

C:

Both steps target different results—allocate additional reasoning effort to a complex multistep decision where supported and reject incomplete or malformed response data before business logic consumes it—because reasoning settings can change token consumption and response latency and should be evaluated for the particular workload and JSON parsing validates syntax, while separate schema and semantic checks verify types and permitted values; the required capabilities are instead to stop interpreting non-2xx API responses as normal assistant messages and to protect model API traffic against an impersonated network endpoint.

D:

The first step is appropriate for the requirement 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 other step instead aims to reject incomplete or malformed response data before business logic consumes it because JSON parsing validates syntax, while separate schema and semantic checks verify types and permitted values, which does not meet the second requirement to protect model API traffic against an impersonated network endpoint.

Question 355

A team maintaining a software-development agent must meet 2 separate acceptance criteria: (1) build a Claude-powered coding agent with an established runtime for tool orchestration; (2) plan another step only after inspecting the result of a requested external operation. Which two implementation actions directly satisfy the stated requirements? Choose TWO.

  1. Use the Claude Agent SDK for the agent's execution harness.
  2. Delegate source analysis to a subagent and return a concise result.
  3. Alternate model tool-use requests with tool results until completion.
  4. Apply limits for steps, elapsed time, and token or tool expenditure.
  5. Compact prior progress into an accurate working summary.

Explanation

Correct Answers

A:

The test to build a Claude-powered coding agent with an established runtime for tool orchestration is met because the SDK provides structured primitives for agent sessions and tool execution, reducing the custom loop infrastructure the application must maintain; the remaining selection covers the separate need to plan another step only after inspecting the result of a requested external operation.

C:

The test to plan another step only after inspecting the result of a requested external operation is met because the loop returns observations to the model so subsequent decisions are informed by actual tool output rather than imagined execution; the remaining selection covers the separate need to build a Claude-powered coding agent with an established runtime for tool orchestration.

Incorrect Answers

B:

The action is useful for the separate capability to keep a large source-reading task from crowding the coordinator's working context because isolated workspaces let a specialist process extensive material while the parent receives only task-relevant evidence instead of every intermediate token; however, the required test outcomes are to build a Claude-powered coding agent with an established runtime for tool orchestration and to plan another step only after inspecting the result of a requested external operation.

D:

The action is useful for the separate capability to impose predictable maximum cost and wall-clock exposure on autonomous execution because runtime budgets are external constraints on agent activity and must be enforced by the host, not inferred from a cooperative natural-language instruction; however, the required test outcomes are to build a Claude-powered coding agent with an established runtime for tool orchestration and to plan another step only after inspecting the result of a requested external operation.

E:

The action is useful for the separate capability to continue a long agent run after intermediate observations dominate the context because compaction preserves decisions and unresolved work while reducing token pressure; losing constraints in the summary can still cause errors; however, the required test outcomes are to build a Claude-powered coding agent with an established runtime for tool orchestration and to plan another step only after inspecting the result of a requested external operation.

Question 356

An architecture decision for a TypeScript Claude integration must support explain why a previously cached prompt is now charged as new input and ask Claude to interpret a supported image supplied as part of a request. Which option gives a complete rather than partial solution?

  1. Perform both steps: (1) Inspect changes to content and ordering before the cache breakpoint; (2) assign and preserve unique custom identifiers for batched requests.
  2. Perform both steps: (1) Inspect changes to content and ordering before the cache breakpoint; (2) include a typed image content block in the message.
  3. Perform both steps: (1) Branch on stop_reason tool_use and dispatch the declared calls; (2) assign and preserve unique custom identifiers for batched requests.
  4. Perform both steps: (1) Branch on stop_reason tool_use and dispatch the declared calls; (2) include a typed image content block in the message.

Explanation

Correct Answer

B:

The first change enables the team to explain why a previously cached prompt is now charged as new input because cache reuse depends on matching prefix material; adding text to the front changes what is cached even if later text is unchanged; the second addresses the separate need to ask Claude to interpret a supported image supplied as part of a request because multimodal input must use the API's image representation and supported sourcing; merely writing a filename in text does not transfer pixels.

Incorrect Answers

A:

The first step is appropriate for the requirement to explain why a previously cached prompt is now charged as new input because cache reuse depends on matching prefix material; adding text to the front changes what is cached even if later text is unchanged; the other step instead aims 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, which does not meet the second requirement to ask Claude to interpret a supported image supplied as part of a request.

C:

Both steps target different results—recognize that the model is asking the application to execute a client tool and map out-of-order batch outputs back to source records reliably—because a tool-use stop requires host-side handling and a follow-up message containing results, not presenting the call as final prose and asynchronous results need stable correlation keys; relying on processing order can attach an output to the wrong input; the required capabilities are instead to explain why a previously cached prompt is now charged as new input and to ask Claude to interpret a supported image supplied as part of a request.

D:

The second step meets the requirement to ask Claude to interpret a supported image supplied as part of a request because multimodal input must use the API's image representation and supported sourcing; merely writing a filename in text does not transfer pixels; the first step would only help 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, leaving the separate requirement to explain why a previously cached prompt is now charged as new input unmet.

Question 357

An architecture review of a knowledge-grounded assistant asks how to stop verbose tool payloads from consuming most of a long-running agent's context. Which action is the best fit?

  1. Reserve an output budget while enforcing input-context limits.
  2. Minimize or redact sensitive fields before constructing context.
  3. Retain essential fields and discard irrelevant tool output from active context.
  4. Recheck current plans against a stored objective and constraints.

Explanation

Correct Answer

C:

To stop verbose tool payloads from consuming most of a long-running agent's context in this knowledge-grounded assistant, summarizing or filtering observations keeps the reasoning window focused without deleting required facts from durable application state; the proposed implementation directly supplies that capability.

Incorrect Answers

A:

The proposed alternative helps to leave space for a model to generate a complete answer after adding source documents because input instructions, history, and evidence share the overall model capacity with generation, so filling it entirely can truncate output; that mechanism does not meet the different requirement to stop verbose tool payloads from consuming most of a long-running agent's context.

B:

The proposed alternative helps to provide a document's necessary meaning without disclosing irrelevant personal identifiers because data minimization reduces the sensitive information transmitted for inference while retaining the facts the task legitimately requires; that mechanism does not meet the different requirement to stop verbose tool payloads from consuming most of a long-running agent's context.

D:

The proposed alternative helps to identify when a long agent discussion begins solving a different problem than the user approved because explicit objective comparison detects drift before tool actions diverge from the original task; extra history alone cannot guarantee alignment; that mechanism does not meet the different requirement to stop verbose tool payloads from consuming most of a long-running agent's context.

Question 358

For an operations research assistant, the missing capability is to avoid adding orchestration infrastructure to a single request-and-response feature. What should the developers implement?

  1. Prefer a simple API call over a general-purpose multi-agent framework.
  2. Run independent branches concurrently and merge their outputs afterward.
  3. Use a supervisor agent to delegate bounded tasks and synthesize results.
  4. Write selected durable facts to a controlled external memory store.

Explanation

Correct Answer

A:

To avoid adding orchestration infrastructure to a single request-and-response feature in this operations research assistant, frameworks are useful for genuinely branching, long-lived execution, but increase complexity when no orchestration problem exists; the proposed implementation directly supplies that capability.

Incorrect Answers

B:

The proposed alternative helps to reduce elapsed time for independent extraction tasks that do not share mutable state because parallel orchestration shortens the critical path when calls have no dependencies, while the merger checks completeness before proceeding; that mechanism does not meet the different requirement to avoid adding orchestration infrastructure to a single request-and-response feature.

C:

The proposed alternative helps to coordinate several specialized investigations and assemble one accountable final answer because a supervisor owns the overall objective and integrates subagent findings; each delegate should have a narrow mandate and return a constrained summary; that mechanism does not meet the different requirement to avoid adding orchestration infrastructure to a single request-and-response feature.

D:

The proposed alternative helps to preserve durable user-approved facts across agent sessions outside the context window because external memory allows selective retrieval in later sessions, unlike temporary model context that disappears when calls end; that mechanism does not meet the different requirement to avoid adding orchestration infrastructure to a single request-and-response feature.

Question 359

A design review for a model selection trial asks for support to choose between writing raw REST requests and using an official client library, together with a way to investigate inconsistent throughput during a traffic spike. Select the two appropriate changes. Choose TWO.

  1. Prefer a maintained SDK when its typed methods and retry behavior fit the integration.
  2. Use a WebSocket protocol only when the target endpoint actually offers one.
  3. Record rate-limit status and relevant response headers alongside request metrics.
  4. Use a synchronous client with explicit timeout and error handling.
  5. Propagate cancellation to the active network request when supported.

Explanation

Correct Answers

A:

To choose between writing raw REST requests and using an official client library as part of a design that must also investigate inconsistent throughput during a traffic spike, sDKs wrap the same underlying service API and can simplify serialization, error types, and streaming without removing protocol constraints; this is why the action is one of the two required selections.

C:

To investigate inconsistent throughput during a traffic spike as part of a design that must also choose between writing raw REST requests and using an official client library, quota-related measurements explain capacity failures that may otherwise look like randomly slow model inference; this is why the action is one of the two required selections.

Incorrect Answers

B:

This alternative serves the separate goal to build bidirectional persistent event communication where the chosen platform supports it because webSockets can carry two-way events over a long-lived connection; they are not a drop-in replacement for every HTTP or SSE endpoint; the required outcomes are instead to choose between writing raw REST requests and using an official client library and to investigate inconsistent throughput during a traffic spike.

D:

This alternative serves the separate goal to implement a low-volume command-line utility that handles one inference at a time because synchronous calls are simplest for strictly sequential work, where concurrency would add complexity without improving throughput; the required outcomes are instead to choose between writing raw REST requests and using an official client library and to investigate inconsistent throughput during a traffic spike.

E:

This alternative serves the separate goal to stop a no-longer-needed inference when the user's request is cancelled because cancelling work can free local resources and avoid processing obsolete results, though already-incurred provider work may still be billed; the required outcomes are instead to choose between writing raw REST requests and using an official client library and to investigate inconsistent throughput during a traffic spike.

Question 360

A team maintaining an internal agent-extension rollout has two different engineering requirements: (1) supply a read-only policy manual to an MCP client without giving it an action endpoint, and (2) obtain a capability already provided as a supported Anthropic server tool. Which TWO independent changes directly meet them? Choose TWO.

  1. Use the MCP stdio transport with properly framed protocol messages.
  2. Host a restricted client tool in the private application environment.
  3. Expose the manual as an MCP resource rather than a tool.
  4. Implement the appropriate MCP capability listing and metadata endpoints.
  5. Configure the built-in capability instead of recreating it as a local handler.

Explanation

Correct Answers

C:

To supply a read-only policy manual to an MCP client without giving it an action endpoint as part of a design that must also obtain a capability already provided as a supported Anthropic server tool, a resource is contextual data; representing a passive document as an imperative function adds unnecessary execution semantics; this is why the action is one of the two required selections.

E:

To obtain a capability already provided as a supported Anthropic server tool as part of a design that must also supply a read-only policy manual to an MCP client without giving it an action endpoint, provider tools simplify execution ownership, though availability, cost, and data handling still require assessment; this is why the action is one of the two required selections.

Incorrect Answers

A:

This alternative serves the separate goal to launch a local MCP server as a subprocess managed by a compatible client because standard input and output suit local process integration, where protocol output must not be mixed with arbitrary logging noise; the required outcomes are instead to supply a read-only policy manual to an MCP client without giving it an action endpoint and to obtain a capability already provided as a supported Anthropic server tool.

B:

This alternative serves the separate goal to execute a high-privilege enterprise action that must remain inside a corporate network because a private handler provides local network and IAM controls that a generic provider-executed server tool may not possess; the required outcomes are instead to supply a read-only policy manual to an MCP client without giving it an action endpoint and to obtain a capability already provided as a supported Anthropic server tool.

D:

This alternative serves the separate goal to help a client learn which callable and informational capabilities a server offers because discovery advertises tools, resources, or prompts so clients can use supported contracts rather than guessing names or arguments; the required outcomes are instead to supply a read-only policy manual to an MCP client without giving it an action endpoint and to obtain a capability already provided as a supported Anthropic server tool.

Claude Certified Developer - Foundations (CCDV-F) Practice Tests