Anthropic CCDV-F (Claude Certified Developer - Foundations) Practice Test 38

Question 370

For a secure tool-execution boundary, the missing capability is to stop frontend users from receiving unrestricted backend tool credentials. What should the developers implement?

  1. Validate or moderate final content before release.
  2. Check resource-level authorization for every sensitive tool request.
  3. Use a server-side workload identity or scoped service credential.
  4. Remove or mask unnecessary personal information before model submission.

Explanation

Correct Answer

C:

To stop frontend users from receiving unrestricted backend tool credentials in this secure tool-execution boundary, service identities tie privileges to a runtime workload and avoid exposing long-lived secrets to browsers or untrusted users; the proposed implementation directly supplies that capability.

Incorrect Answers

A:

The proposed alternative helps to prevent generated advice that violates the product's publication rules from being shown because output checks catch issues emerging during generation; they complement rather than replace input controls and tool restrictions; that mechanism does not meet the different requirement to stop frontend users from receiving unrestricted backend tool credentials.

B:

The proposed alternative helps 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; that mechanism does not meet the different requirement to stop frontend users from receiving unrestricted backend tool credentials.

D:

The proposed alternative helps to protect customer identity fields when only document topics are needed for classification because minimization limits exposure while preserving task inputs; relying on the model to ignore visible PII does not provide the same protection; that mechanism does not meet the different requirement to stop frontend users from receiving unrestricted backend tool credentials.

Question 371

While planning a model selection trial, the product owner specifies that it must preserve arrays of typed content blocks when sending multimodal input. Which solution fulfills the requirement?

  1. Measure or count input and output tokens with the supported tokenizer or API.
  2. Run contract and integration tests against the upgraded SDK in staging.
  3. Serialize the documented nested request shape rather than flattening it.
  4. Keep HTTPS certificate and hostname verification enabled.

Explanation

Correct Answer

C:

To preserve arrays of typed content blocks when sending multimodal input in this model selection trial, typed content and role fields carry protocol meaning that is lost if developers simply concatenate them into plain text; the proposed implementation directly supplies that capability.

Incorrect Answers

A:

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 preserve arrays of typed content blocks when sending multimodal input.

B:

The proposed alternative helps to detect changed library behavior before rolling a new version into production because client-library version shifts can affect serialization or errors even without changes to business prompts, so they deserve release validation; that mechanism does not meet the different requirement to preserve arrays of typed content blocks when sending multimodal input.

D:

The proposed alternative helps 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; that mechanism does not meet the different requirement to preserve arrays of typed content blocks when sending multimodal input.

Question 372

A planned upgrade to a repository-level Claude Code rollout has two requirements that must both pass: continue a previously interrupted Claude Code work session; run a Claude Code task from an automated CI job without an interactive terminal. Identify the two direct implementations. Choose TWO.

  1. Configure a specialized Claude Code subagent with a narrow remit.
  2. Invoke the supported noninteractive or print-mode command interface.
  3. Place a path-targeted rule in the repository's rules configuration.
  4. Initialize project guidance and review the resulting CLAUDE.md content.
  5. Use the supported session continuation or resume mechanism.

Explanation

Correct Answers

B:

To run a Claude Code task from an automated CI job without an interactive terminal as part of a design that must also continue a previously interrupted Claude Code work session, headless operation outputs results for automation and requires explicit credential, permission, and failure handling in the CI environment; this is why the action is one of the two required selections.

E:

To continue a previously interrupted Claude Code work session as part of a design that must also run a Claude Code task from an automated CI job without an interactive terminal, restoring session context reduces lost work while preserving continuity of prior tool activity; opening a new unrelated session does not do the same; this is why the action is one of the two required selections.

Incorrect Answers

A:

This alternative serves the separate goal to delegate an independent code review without filling the main coding session with details because subagents can focus on a task and return a summary, limiting noise in the parent session and allowing separate permissions when configured; the required outcomes are instead to continue a previously interrupted Claude Code work session and to run a Claude Code task from an automated CI job without an interactive terminal.

C:

This alternative serves the separate goal to apply a coding convention only when Claude Code edits a particular path because scoped rules keep specialized instructions close to their applicable files, avoiding global instructions that conflict with other modules; the required outcomes are instead to continue a previously interrupted Claude Code work session and to run a Claude Code task from an automated CI job without an interactive terminal.

D:

This alternative serves the separate goal to introduce Claude Code to an unfamiliar codebase's conventions and commands because repository setup discovers useful project practices, but autogenerated guidance still needs a human check for accuracy and unnecessary detail; the required outcomes are instead to continue a previously interrupted Claude Code work session and to run a Claude Code task from an automated CI job without an interactive terminal.

Question 373

The team responsible for a multi-tenant support application needs to avoid retrying a malformed request that the API consistently rejects. What is the most direct engineering change?

  1. Set the appropriate application/json content type and serialize a valid payload.
  2. Classify permanent client errors separately from transient failures.
  3. Log status, timing, and correlation IDs with sensitive fields redacted.
  4. Give mutating operations an application-level idempotency key.

Explanation

Correct Answer

B:

To avoid retrying a malformed request that the API consistently rejects in this multi-tenant support application, a deterministic validation or authorization error requires fixing the request or credential, not another identical attempt; the proposed implementation directly supplies that capability.

Incorrect Answers

A:

The proposed alternative helps to have the server parse a JSON request body predictably because a JSON body requires both correct serialization and matching HTTP metadata; raw language-model text cannot replace a valid request object; that mechanism does not meet the different requirement to avoid retrying a malformed request that the API consistently rejects.

C:

The proposed alternative helps to debug a failed request without exposing the user's confidential document because operational records can identify a failure path without retaining raw secrets or protected prompt content; that mechanism does not meet the different requirement to avoid retrying a malformed request that the API consistently rejects.

D:

The proposed alternative helps to prevent duplicate purchases or tickets if an agent repeats a tool call because idempotency ensures replays of the same intended action do not create duplicate real-world effects even when orchestration retries; that mechanism does not meet the different requirement to avoid retrying a malformed request that the API consistently rejects.

Question 374

The technical owner of a token-consumption optimization sprint has to allow supported models to vary internal reasoning effort according to task difficulty. A separate requirement is to allocate additional reasoning effort to a complex multistep decision where supported. Which two-part remediation covers both?

  1. Perform both steps: (1) Measure or count input and output tokens with the supported tokenizer or API; (2) use the model's documented thinking capability within the latency budget.
  2. Perform both steps: (1) Configure supported adaptive thinking behavior rather than forcing a fixed pattern; (2) include several carefully selected and diverse demonstrations.
  3. Perform both steps: (1) Configure supported adaptive thinking behavior rather than forcing a fixed pattern; (2) use the model's documented thinking capability within the latency budget.
  4. Perform both steps: (1) Measure or count input and output tokens with the supported tokenizer or API; (2) include several carefully selected and diverse demonstrations.

Explanation

Correct Answer

C:

The first change enables the team 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; the second addresses the separate need 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.

Incorrect Answers

A:

The second step meets the requirement 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; the first step would only help 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, leaving the separate requirement to allow supported models to vary internal reasoning effort according to task difficulty unmet.

B:

The first step is appropriate for the requirement 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; the other step instead aims to reduce inconsistent labeling across nuanced examples because multiple examples communicate decision boundaries and desired formatting, although irrelevant or misleading demonstrations can bias the model, which does not meet the second requirement to allocate additional reasoning effort to a complex multistep decision where supported.

D:

Both steps target different results—estimate model usage from encoded text rather than counting words and reduce inconsistent labeling across nuanced examples—because billing and context consumption track tokens, which do not map one-to-one to English words or characters and multiple examples communicate decision boundaries and desired formatting, although irrelevant or misleading demonstrations can bias the model; the required capabilities are instead to allow supported models to vary internal reasoning effort according to task difficulty and to allocate additional reasoning effort to a complex multistep decision where supported.

Question 375

An architecture review of a tool-calling Claude workflow asks how to enforce a mandatory fixed approval step regardless of what instructions the model follows. Which action is the best fit?

  1. Expose the manual as an MCP resource rather than a tool.
  2. Implement the approval in application workflow control rather than only in a Skill.
  3. Expose the material as an MCP resource with a stable identifier.
  4. Package the connector behind an MCP server.

Explanation

Correct Answer

B:

To enforce a mandatory fixed approval step regardless of what instructions the model follows in this tool-calling Claude workflow, skills can guide agent behavior, but deterministic policy gates must live in trusted code so they cannot be skipped by model interpretation; the proposed implementation directly supplies that capability.

Incorrect Answers

A:

The proposed alternative helps 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; that mechanism does not meet the different requirement to enforce a mandatory fixed approval step regardless of what instructions the model follows.

C:

The proposed alternative helps to make reference material available for clients to read as context because resources represent consumable data, contrasting with tools that invoke an operation or prompts that package reusable instructions; that mechanism does not meet the different requirement to enforce a mandatory fixed approval step regardless of what instructions the model follows.

D:

The proposed alternative helps to share one integration with several compatible AI clients or agent runtimes because MCP standardizes capability exposure across clients, reducing duplicate integration logic compared with proprietary tool schemas everywhere; that mechanism does not meet the different requirement to enforce a mandatory fixed approval step regardless of what instructions the model follows.

Question 376

For a secure tool-execution boundary, the missing capability is to reject a malicious instruction inserted into a web-search result rather than the user's direct request. What should the developers implement?

  1. Apply an input-side policy gate and route flagged cases to a safe path.
  2. Combine prompt boundaries, scoped credentials, validation, and approval gates.
  3. Isolate external search results and validate any resulting tool action independently.
  4. Validate or moderate final content before release.

Explanation

Correct Answer

C:

To reject a malicious instruction inserted into a web-search result rather than the user's direct request in this secure tool-execution boundary, indirect attacks exploit lower-trust retrieval surfaces, so trusting search snippets as operational directives crosses a security boundary; the proposed implementation directly supplies that capability.

Incorrect Answers

A:

The proposed alternative helps to identify prohibited requests before an automated agent invokes business tools because early filtering reduces inappropriate work, but later checks remain necessary because tool outputs can also introduce risks; that mechanism does not meet the different requirement to reject a malicious instruction inserted into a web-search result rather than the user's direct request.

B:

The proposed alternative helps to secure an agent against both malicious retrieved text and unsafe tool execution because defense in depth assumes individual filters can fail; independent enforcement boundaries reduce the chance of harmful side effects; that mechanism does not meet the different requirement to reject a malicious instruction inserted into a web-search result rather than the user's direct request.

D:

The proposed alternative helps to prevent generated advice that violates the product's publication rules from being shown because output checks catch issues emerging during generation; they complement rather than replace input controls and tool restrictions; that mechanism does not meet the different requirement to reject a malicious instruction inserted into a web-search result rather than the user's direct request.

Question 377

While planning a long-context summarization workflow, the product owner specifies that it must measure whether a changed instruction actually improves extraction accuracy. Which solution fulfills the requirement?

  1. Use properly labeled example inputs and outputs instead of mixing speakers.
  2. Have a scoped subagent process sources and return a concise verified handoff.
  3. Hold evaluation inputs constant and compare prompt versions.
  4. Supply representative labeled examples including edge conditions.

Explanation

Correct Answer

C:

To measure whether a changed instruction actually improves extraction accuracy in this long-context summarization workflow, a reproducible prompt test isolates the change being studied; anecdotal output from unrelated cases does not show improvement; the proposed implementation directly supplies that capability.

Incorrect Answers

A:

The proposed alternative helps to maintain label semantics when a few-shot prompt contains dialogue because correct role boundaries keep demonstrations interpretable and prevent an example answer from looking like a live instruction; that mechanism does not meet the different requirement to measure whether a changed instruction actually improves extraction accuracy.

B:

The proposed alternative helps to analyze a large source set without exposing every raw document to the parent agent because separate contexts let specialists work deeply while the coordinator sees only claims and references needed for the final decision; that mechanism does not meet the different requirement to measure whether a changed instruction actually improves extraction accuracy.

D:

The proposed alternative helps to teach a classifier how to label borderline requests consistently because diverse demonstrations make the intended decision boundary clearer than an isolated easy example, provided labels are accurate; that mechanism does not meet the different requirement to measure whether a changed instruction actually improves extraction accuracy.

Question 378

A change request for a high-volume Messages API gateway combines two needs: catch mismatches between model calls, tool handlers, and application persistence, and limit the blast radius of a new model configuration during rollout. Which implementation pair addresses both, not just one?

  1. Perform both steps: (1) Exercise complete request-to-tool-to-response paths in a controlled environment; (2) release to a small monitored cohort before broad deployment.
  2. Perform both steps: (1) Exercise complete request-to-tool-to-response paths in a controlled environment; (2) keep versioned artifacts and a tested rollback procedure.
  3. Perform both steps: (1) Require peer review focused on API, security, and tool-side effects; (2) release to a small monitored cohort before broad deployment.
  4. Perform both steps: (1) Require peer review focused on API, security, and tool-side effects; (2) keep versioned artifacts and a tested rollback procedure.

Explanation

Correct Answer

A:

The first change enables the team to catch mismatches between model calls, tool handlers, and application persistence because unit tests cannot reveal every serialization or permission issue across service boundaries; integration tests validate the connected contract; the second addresses the separate need to limit the blast radius of a new model configuration during rollout because canary exposure provides real traffic signals with an option to stop or roll back, unlike a simultaneous change for every customer.

Incorrect Answers

B:

The first step is appropriate for the requirement to catch mismatches between model calls, tool handlers, and application persistence because unit tests cannot reveal every serialization or permission issue across service boundaries; integration tests validate the connected contract; the other step instead aims to restore a known-working prompt and model configuration after quality degrades because model IDs, prompts, schemas, and tool contracts are release components; restoring just the application binary may not restore prior behavior, which does not meet the second requirement to limit the blast radius of a new model configuration during rollout.

C:

The second step meets the requirement to limit the blast radius of a new model configuration during rollout because canary exposure provides real traffic signals with an option to stop or roll back, unlike a simultaneous change for every customer; the first step would only help to catch unsafe credentials and incorrect tool calls before deployment because human review should check trust boundaries and external actions in addition to style; successful compilation cannot prove these behaviors are safe, leaving the separate requirement to catch mismatches between model calls, tool handlers, and application persistence unmet.

D:

Both steps target different results—catch unsafe credentials and incorrect tool calls before deployment and restore a known-working prompt and model configuration after quality degrades—because human review should check trust boundaries and external actions in addition to style; successful compilation cannot prove these behaviors are safe and model IDs, prompts, schemas, and tool contracts are release components; restoring just the application binary may not restore prior behavior; the required capabilities are instead to catch mismatches between model calls, tool handlers, and application persistence and to limit the blast radius of a new model configuration during rollout.

Question 379

The technical owner of a shared tool gateway for Claude has to execute a high-privilege enterprise action that must remain inside a corporate network. A separate requirement is to maintain a reusable collection of related integrations and agent instructions across projects. Which two-part remediation covers both?

  1. Perform both steps: (1) Implement a client-side handler and return its result with the tool-use identifier; (2) package the compatible components as an installable plugin when supported.
  2. Perform both steps: (1) Host a restricted client tool in the private application environment; (2) package the compatible components as an installable plugin when supported.
  3. Perform both steps: (1) Host a restricted client tool in the private application environment; (2) create a skill containing the procedure and supporting instructions.
  4. Perform both steps: (1) Implement a client-side handler and return its result with the tool-use identifier; (2) create a skill containing the procedure and supporting instructions.

Explanation

Correct Answer

B:

The first change enables the team 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 second addresses the separate need to maintain a reusable collection of related integrations and agent instructions across projects because a plugin groups capabilities for distribution and configuration; a single direct tool is simpler when packaging and reuse are unnecessary.

Incorrect Answers

A:

The second step meets the requirement to maintain a reusable collection of related integrations and agent instructions across projects because a plugin groups capabilities for distribution and configuration; a single direct tool is simpler when packaging and reuse are unnecessary; the first step would only help to connect a Claude tool request to an internal service whose logic the application owns because a declared client tool does not execute itself; application code runs it and feeds results back into the conversation, leaving the separate requirement to execute a high-privilege enterprise action that must remain inside a corporate network unmet.

C:

The first step is appropriate for the requirement 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 other step instead aims 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, which does not meet the second requirement to maintain a reusable collection of related integrations and agent instructions across projects.

D:

Both steps target different results—connect a Claude tool request to an internal service whose logic the application owns and make a repeatable multi-step code-review methodology available to an agent—because a declared client tool does not execute itself; application code runs it and feeds results back into the conversation and a Skill teaches how to perform a task, whereas a tool is the executable capability that performs an external operation; the required capabilities are instead to execute a high-privilege enterprise action that must remain inside a corporate network and to maintain a reusable collection of related integrations and agent instructions across projects.

Question 380

A team maintaining an incident involving failed Claude completions has two different engineering requirements: (1) identify why a request is rejected before any inference happens, and (2) locate which component of a multi-tool answer introduced the wrong value. Which TWO independent changes directly meet them? Choose TWO.

  1. Examine correlated request, tool, and transformation traces stage by stage.
  2. Trace the API credential and access permissions for the 401 or 403 response.
  3. Compare versioned prompts and representative eval cases against model responses.
  4. Inspect stop_reason and output limits before changing the parser.
  5. Read the 400 error details and validate the request payload.

Explanation

Correct Answers

A:

To locate which component of a multi-tool answer introduced the wrong value as part of a design that must also identify why a request is rejected before any inference happens, structured traces reveal whether retrieval, tool execution, postprocessing, or generation caused the defect; reading only the final text cannot isolate it; this is why the action is one of the two required selections.

B:

To identify why a request is rejected before any inference happens as part of a design that must also locate which component of a multi-tool answer introduced the wrong value, unauthorized access fails at the transport or authorization boundary, not because Claude misunderstood the prompt; this is why the action is one of the two required selections.

Incorrect Answers

C:

This alternative serves the separate goal to distinguish a syntactically valid but low-quality answer from a transport outage because a successful HTTP call only proves the protocol worked; quality regressions require output-level evidence and controlled comparisons; the required outcomes are instead to identify why a request is rejected before any inference happens and to locate which component of a multi-tool answer introduced the wrong value.

D:

This alternative serves the separate goal to find why a JSON record unexpectedly ends midway through a field because a max_tokens stop indicates incomplete generation; the parser may be working correctly against a response that was never finished; the required outcomes are instead to identify why a request is rejected before any inference happens and to locate which component of a multi-tool answer introduced the wrong value.

E:

This alternative serves the separate goal to find a request schema error instead of repeatedly changing the system prompt because an invalid API field or incompatible parameter cannot be fixed by natural-language instructions within the malformed request; the required outcomes are instead to identify why a request is rejected before any inference happens and to locate which component of a multi-tool answer introduced the wrong value.

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