Anthropic CCA-F (Claude Certified Architect - Foundations (CCAR-F)) Practice Test 21

Question 200

Tests for a continuous integration repair assistant will pass only if two nested packages follow distinct framework conventions and if a shared skill can take a supplied target name or file path. Which proposed design directly satisfies both checks?

  1. create separate path-specific rule files for each package; and define how the skill accepts and uses invocation arguments
  2. write one ambiguous instruction for both packages; and define how the skill accepts and uses invocation arguments
  3. write one ambiguous instruction for both packages; and copy a separate static instruction file for every target path
  4. create separate path-specific rule files for each package; and copy a separate static instruction file for every target path

Explanation

Correct Answer

A:

In a continuous integration repair assistant, scoped conventions avoid contradictory patterns between codebases sharing a monorepo; alongside this, parameterized workflows apply one reviewed procedure to different resources without hardcoding targets, satisfying the distinct requirements that two nested packages follow distinct framework conventions and that a shared skill can take a supplied target name or file path.

Incorrect Answers

B:

In a continuous integration repair assistant, parameterized workflows apply one reviewed procedure to different resources without hardcoding targets, but the model must guess which framework convention applies; therefore the separate requirement that two nested packages follow distinct framework conventions would remain unsatisfied.

C:

In a continuous integration repair assistant, the model must guess which framework convention applies and duplicated procedures are costly to maintain and can diverge; the design would fail both conditions: two nested packages follow distinct framework conventions and a shared skill can take a supplied target name or file path.

D:

In a continuous integration repair assistant, scoped conventions avoid contradictory patterns between codebases sharing a monorepo, but duplicated procedures are costly to maintain and can diverge; therefore the separate requirement that a shared skill can take a supplied target name or file path would remain unsatisfied.

Question 201

An architecture review of a multi-source competitor research system finds two missing guarantees: a report separates documented findings from its own inference; a failed MCP operation is visible as an error to the client. What two-part change closes both gaps?

  1. present every deduction as a direct source quotation; and return a success-shaped value saying the operation probably worked
  2. label direct source facts separately from analytical conclusions; and return a success-shaped value saying the operation probably worked
  3. label direct source facts separately from analytical conclusions; and mark the tool result as an error and supply structured failure details
  4. present every deduction as a direct source quotation; and mark the tool result as an error and supply structured failure details

Explanation

Correct Answer

C:

In a multi-source competitor research system, distinguishing evidence from interpretation prevents inference being misattributed to a document; alongside this, explicit error status separates failed execution from valid data content, satisfying the distinct requirements that a report separates documented findings from its own inference and that a failed MCP operation is visible as an error to the client.

Incorrect Answers

A:

In a multi-source competitor research system, unsupported attribution misrepresents what the source actually says and downstream agents can mistake a failure for verified business data; the design would fail both conditions: a report separates documented findings from its own inference and a failed MCP operation is visible as an error to the client.

B:

In a multi-source competitor research system, distinguishing evidence from interpretation prevents inference being misattributed to a document, but downstream agents can mistake a failure for verified business data; therefore the separate requirement that a failed MCP operation is visible as an error to the client would remain unsatisfied.

D:

In a multi-source competitor research system, explicit error status separates failed execution from valid data content, but unsupported attribution misrepresents what the source actually says; therefore the separate requirement that a report separates documented findings from its own inference would remain unsatisfied.

Question 202

In a shared-library improvement effort, the design must ensure that a shared skill can take a supplied target name or file path and that the assistant locates files containing a particular symbol or string. Choose TWO independent changes that directly meet these two requirements.

  1. use Grep to search file contents
  2. define how the skill accepts and uses invocation arguments
  3. copy a separate static instruction file for every target path
  4. scan only filenames with Glob

Explanation

Correct Answers

A:

For a shared-library improvement effort, text search locates occurrences efficiently before focused reading, which directly establishes that the assistant locates files containing a particular symbol or string; the separate requirement that a shared skill can take a supplied target name or file path is handled by a different action.

B:

For a shared-library improvement effort, parameterized workflows apply one reviewed procedure to different resources without hardcoding targets, which directly establishes that a shared skill can take a supplied target name or file path; the separate requirement that the assistant locates files containing a particular symbol or string is handled by a different action.

Incorrect Answers

C:

For a shared-library improvement effort, duplicated procedures are costly to maintain and can diverge, so this option cannot establish that a shared skill can take a supplied target name or file path; correctly addressing the assistant locates files containing a particular symbol or string elsewhere does not repair that deficiency.

D:

For a shared-library improvement effort, path matching does not inspect symbol occurrences inside files, so this option cannot establish that the assistant locates files containing a particular symbol or string; correctly addressing a shared skill can take a supplied target name or file path elsewhere does not repair that deficiency.

Question 203

An architecture review of a repository release validation process finds two missing guarantees: a team compares two candidate patches against the same baseline; a forced tool call still honors business authorization. What two-part change closes both gaps?

  1. implement only evaluate alternatives using the same tests and scoped diffs; make no corresponding change for the second requirement
  2. choose whichever patch has the more persuasive natural-language summary; and treat a forced tool choice as equivalent to user approval
  3. choose whichever patch has the more persuasive natural-language summary; and combine tool_choice with server-side policy checks
  4. evaluate alternatives using the same tests and scoped diffs; and treat a forced tool choice as equivalent to user approval
  5. evaluate alternatives using the same tests and scoped diffs; and combine tool_choice with server-side policy checks

Explanation

Correct Answer

E:

In a repository release validation process, consistent acceptance evidence reveals which candidate improves the relevant behavior; alongside this, selection controls when a tool is proposed but does not grant permission to execute, satisfying the distinct requirements that a team compares two candidate patches against the same baseline and that a forced tool call still honors business authorization.

Incorrect Answers

A:

In a repository release validation process, consistent acceptance evidence reveals which candidate improves the relevant behavior, but the independent mechanism for the other criterion is absent: selection controls when a tool is proposed but does not grant permission to execute; as written, the design does not ensure that a forced tool call still honors business authorization.

B:

In a repository release validation process, rhetorical fluency does not establish code correctness and tool selection and authorization are different controls; the design would fail both conditions: a team compares two candidate patches against the same baseline and a forced tool call still honors business authorization.

C:

In a repository release validation process, selection controls when a tool is proposed but does not grant permission to execute, but rhetorical fluency does not establish code correctness; therefore the separate requirement that a team compares two candidate patches against the same baseline would remain unsatisfied.

D:

In a repository release validation process, consistent acceptance evidence reveals which candidate improves the relevant behavior, but tool selection and authorization are different controls; therefore the separate requirement that a forced tool call still honors business authorization would remain unsatisfied.

Question 204

Production readiness for a supplier invoice extractor depends on ensuring that one specialist's crash does not discard other verified outputs and that persistent ambiguous data receives human review. Select the most appropriate combined approach.

  1. isolate branch failures and preserve completed artifacts for coordination; and keep retrying until the model finally emits something that passes
  2. restart every agent and erase prior results after a single crash; and keep retrying until the model finally emits something that passes
  3. restart every agent and erase prior results after a single crash; and stop after defined retries and route unresolved cases to a reviewer
  4. isolate branch failures and preserve completed artifacts for coordination; and stop after defined retries and route unresolved cases to a reviewer

Explanation

Correct Answer

D:

In a supplier invoice extractor, fault containment avoids losing good work while a failed branch is repaired; alongside this, a bounded escalation path prevents infinite loops and invented resolutions, satisfying the distinct requirements that one specialist's crash does not discard other verified outputs and that persistent ambiguous data receives human review.

Incorrect Answers

A:

In a supplier invoice extractor, fault containment avoids losing good work while a failed branch is repaired, but unlimited retries waste resources and can reinforce guesses; therefore the separate requirement that persistent ambiguous data receives human review would remain unsatisfied.

B:

In a supplier invoice extractor, unaffected branches would be redone and provenance could be lost and unlimited retries waste resources and can reinforce guesses; the design would fail both conditions: one specialist's crash does not discard other verified outputs and persistent ambiguous data receives human review.

C:

In a supplier invoice extractor, a bounded escalation path prevents infinite loops and invented resolutions, but unaffected branches would be redone and provenance could be lost; therefore the separate requirement that one specialist's crash does not discard other verified outputs would remain unsatisfied.

Question 205

While preparing a structured report conversion service for deployment, engineers must ensure that a formatter handles diverse legitimate examples consistently; they must also ensure that each batch result is matched to its originating record. Which combination best satisfies those requirements?

  1. supply only a bare instruction to keep formatting consistent; and assign stable custom IDs and map results by those IDs
  2. give a small set of representative input-output demonstrations; and assign stable custom IDs and map results by those IDs
  3. supply only a bare instruction to keep formatting consistent; and match returned results to input rows only by array position
  4. give a small set of representative input-output demonstrations; and match returned results to input rows only by array position
  5. implement only give a small set of representative input-output demonstrations; make no corresponding change for the second requirement

Explanation

Correct Answer

B:

In a structured report conversion service, few-shot examples show the intended mapping more concretely than broad style adjectives; alongside this, asynchronous completion order need not match submission order, satisfying the distinct requirements that a formatter handles diverse legitimate examples consistently and that each batch result is matched to its originating record.

Incorrect Answers

A:

In a structured report conversion service, asynchronous completion order need not match submission order, but abstract instructions leave many output choices open; therefore the separate requirement that a formatter handles diverse legitimate examples consistently would remain unsatisfied.

C:

In a structured report conversion service, abstract instructions leave many output choices open and out-of-order or failed items can corrupt record association; the design would fail both conditions: a formatter handles diverse legitimate examples consistently and each batch result is matched to its originating record.

D:

In a structured report conversion service, few-shot examples show the intended mapping more concretely than broad style adjectives, but out-of-order or failed items can corrupt record association; therefore the separate requirement that each batch result is matched to its originating record would remain unsatisfied.

E:

In a structured report conversion service, few-shot examples show the intended mapping more concretely than broad style adjectives, but the independent mechanism for the other criterion is absent: asynchronous completion order need not match submission order; as written, the design does not ensure that each batch result is matched to its originating record.

Question 206

Before expanding an internal knowledge research system, the owner must demonstrate that the system can continue when one specialist returns an explicit failure, while also demonstrating that the model can use a tool result in its next decision. Which design provides both controls?

  1. treat every subagent error string as a valid research conclusion; and append the assistant tool-use response and matching user tool_result blocks to history
  2. record per-branch status and let the coordinator choose retry or partial response; and append the assistant tool-use response and matching user tool_result blocks to history
  3. record per-branch status and let the coordinator choose retry or partial response; and put the tool output only in application logs
  4. treat every subagent error string as a valid research conclusion; and put the tool output only in application logs

Explanation

Correct Answer

B:

In an internal knowledge research system, the orchestration layer must distinguish missing evidence from a negative factual finding; alongside this, Claude needs the prior request and the execution output in the conversation for the next inference, satisfying the distinct requirements that the system can continue when one specialist returns an explicit failure and that the model can use a tool result in its next decision.

Incorrect Answers

A:

In an internal knowledge research system, Claude needs the prior request and the execution output in the conversation for the next inference, but error text is operational state rather than evidence about the world; therefore the separate requirement that the system can continue when one specialist returns an explicit failure would remain unsatisfied.

C:

In an internal knowledge research system, the orchestration layer must distinguish missing evidence from a negative factual finding, but unseen logs cannot inform the model's next decision; therefore the separate requirement that the model can use a tool result in its next decision would remain unsatisfied.

D:

In an internal knowledge research system, error text is operational state rather than evidence about the world and unseen logs cannot inform the model's next decision; the design would fail both conditions: the system can continue when one specialist returns an explicit failure and the model can use a tool result in its next decision.

Question 207

In an internal knowledge research system, the immediate design requirement is that specialists return comparable artifacts for aggregation. Which change most directly establishes that behavior?

  1. give subagents a common results contract including claim, evidence, and status
  2. have the coordinator compare claims and sources before synthesis
  3. dispatch the independent branches concurrently through the coordinator
  4. merge narrative responses by concatenating their text

Explanation

Correct Answer

A:

For an internal knowledge research system, a typed handoff lets the coordinator compare like fields and detect missing support; this directly establishes that specialists return comparable artifacts for aggregation instead of relying on an unrelated workflow control.

Incorrect Answers

B:

Although central reconciliation can distinguish evidence disagreement from a formatting difference, that only supports the distinct goal that conflicting subagent findings produce one accountable interpretation, not the requirement that specialists return comparable artifacts for aggregation in an internal knowledge research system.

C:

Although parallel delegation reduces latency when individual investigations do not consume each other's outputs, that only supports the distinct goal that independent research branches can finish without waiting on one another, not the requirement that specialists return comparable artifacts for aggregation in an internal knowledge research system.

D:

In an internal knowledge research system, unstructured concatenation hides provenance and duplicate or conflicting claims; as a result, the stated requirement that specialists return comparable artifacts for aggregation would still not be met.

Question 208

Before expanding an account recovery support process, the owner must demonstrate that a truncated response is not mistaken for successful completion, while also demonstrating that temporary service faults can be retried selectively. Which design provides both controls?

  1. handle max_tokens as truncation rather than end_turn; and include a retryable classification and bounded retry advice
  2. interpret every non-tool stop reason as a successful final answer; and retry every error with the same request indefinitely
  3. handle max_tokens as truncation rather than end_turn; and retry every error with the same request indefinitely
  4. interpret every non-tool stop reason as a successful final answer; and include a retryable classification and bounded retry advice

Explanation

Correct Answer

A:

In an account recovery support process, reaching an output limit is a stop condition but does not establish task completion; alongside this, typed transient failures allow backoff without treating permanent errors the same way, satisfying the distinct requirements that a truncated response is not mistaken for successful completion and that temporary service faults can be retried selectively.

Incorrect Answers

B:

In an account recovery support process, a length-limited result may be incomplete or unusable and permission and validation failures will not heal through repetition; the design would fail both conditions: a truncated response is not mistaken for successful completion and temporary service faults can be retried selectively.

C:

In an account recovery support process, reaching an output limit is a stop condition but does not establish task completion, but permission and validation failures will not heal through repetition; therefore the separate requirement that temporary service faults can be retried selectively would remain unsatisfied.

D:

In an account recovery support process, typed transient failures allow backoff without treating permanent errors the same way, but a length-limited result may be incomplete or unusable; therefore the separate requirement that a truncated response is not mistaken for successful completion would remain unsatisfied.

Question 209

In a pull-request review pipeline, the design must ensure that workload growth does not cause uncontrolled parallel fan-out and that a forced tool call still honors business authorization. Choose TWO independent changes that directly meet these two requirements.

  1. launch all possible child tasks at once without resource controls
  2. treat a forced tool choice as equivalent to user approval
  3. combine tool_choice with server-side policy checks
  4. limit concurrency and schedule independent jobs against execution budgets

Explanation

Correct Answers

C:

For a pull-request review pipeline, selection controls when a tool is proposed but does not grant permission to execute, which directly establishes that a forced tool call still honors business authorization; the separate requirement that workload growth does not cause uncontrolled parallel fan-out is handled by a different action.

D:

For a pull-request review pipeline, bounded orchestration balances throughput with tool quotas and downstream service capacity, which directly establishes that workload growth does not cause uncontrolled parallel fan-out; the separate requirement that a forced tool call still honors business authorization is handled by a different action.

Incorrect Answers

A:

For a pull-request review pipeline, unbounded fan-out can overwhelm tools and blow through cost budgets, so this option cannot establish that workload growth does not cause uncontrolled parallel fan-out; correctly addressing a forced tool call still honors business authorization elsewhere does not repair that deficiency.

B:

For a pull-request review pipeline, tool selection and authorization are different controls, so this option cannot establish that a forced tool call still honors business authorization; correctly addressing workload growth does not cause uncontrolled parallel fan-out elsewhere does not repair that deficiency.

Question 210

Tests for a customer account helpdesk agent will pass only if several requested client tools are returned in one response and if a conversational task does not invoke tools needlessly. Which proposed design directly satisfies both checks?

  1. execute every tool_use block and return a result with each corresponding ID; and use automatic tool choice when a call is optional
  2. execute only the first returned tool call; and use automatic tool choice when a call is optional
  3. execute every tool_use block and return a result with each corresponding ID; and force a tool on every message regardless of intent
  4. execute only the first returned tool call; and force a tool on every message regardless of intent

Explanation

Correct Answer

A:

In a customer account helpdesk agent, a multi-tool assistant turn can request independent actions and must receive all matching outcomes; alongside this, auto allows the model to answer directly when no external side effect or fresh data is needed, satisfying the distinct requirements that several requested client tools are returned in one response and that a conversational task does not invoke tools needlessly.

Incorrect Answers

B:

In a customer account helpdesk agent, auto allows the model to answer directly when no external side effect or fresh data is needed, but the unanswered tool uses remain unresolved and later reasoning loses evidence; therefore the separate requirement that several requested client tools are returned in one response would remain unsatisfied.

C:

In a customer account helpdesk agent, a multi-tool assistant turn can request independent actions and must receive all matching outcomes, but mandatory calls add latency and can invoke irrelevant operations; therefore the separate requirement that a conversational task does not invoke tools needlessly would remain unsatisfied.

D:

In a customer account helpdesk agent, the unanswered tool uses remain unresolved and later reasoning loses evidence and mandatory calls add latency and can invoke irrelevant operations; the design would fail both conditions: several requested client tools are returned in one response and a conversational task does not invoke tools needlessly.

Claude Certified Architect - Foundations (CCAR-F) (CCA-F) Practice Tests