Anthropic CCAR-P (Claude Certified Architect - Professional) Practice Test 45

Question 440

A compliance auditor requests proof that a human approval gate is enforced in production. Which control should govern the first implementation?

  1. For this production path, version the RAG architecture and source lifecycle with reproducible index-build records.
  2. For the intended users, create an operational handoff covering IAM, integrations, secrets, owners and incident escalation.
  3. Within the delivery plan, document the control design and link it to immutable approval-event evidence.
  4. At the system boundary, supply sequence diagrams and failure-mode tables covering ordering, retries and idempotency.
  5. For the primary workstream, maintain a living ADR register with owners, triggers and supersession history.

Explanation

Correct Answer

C:

Architecture assertions need operational artifacts to demonstrate the control actually runs; one operational check should confirm that a sampled transaction trace shows the required authorized approval before commit.

Incorrect Answers

A:

Retrieval behavior cannot be recreated from the generation prompt alone; this applies when a team documents prompts but not which corpus, embedding model and chunking policy produced answers, but here a compliance auditor requests proof that a human approval gate is enforced in production.

B:

A supportable deployment requires responsibility and recovery documentation beyond diagrams; the alternative helps if the delivery team leaves after launch with no record of tool privileges or secret owners, instead of when a compliance auditor requests proof that a human approval gate is enforced in production.

D:

Multi-stage side effects need explicit operational contracts to prevent replay and duplication errors; its scope includes cases where an orchestrated workflow uses multiple asynchronous queues whose retry semantics are undocumented, whereas the problem presented is that a compliance auditor requests proof that a human approval gate is enforced in production.

E:

Decision provenance reduces rediscovery when regulations, usage and models evolve; it targets cases where a solution has dozens of architectural assumptions recorded in chat messages only, but this decision concerns a compliance auditor requests proof that a human approval gate is enforced in production.

Question 441

A service outage requires rollback, but deployment notes contain no previous artifact versions. What intervention best matches the stated boundary?

  1. In the proposed design, include rollback procedures, dependencies and version compatibility in the runbook.
  2. Under the reviewed approach, version the RAG architecture and source lifecycle with reproducible index-build records.
  3. For this production path, document the control design and link it to immutable approval-event evidence.
  4. In the proposed implementation, create an operational handoff covering IAM, integrations, secrets, owners and incident escalation.

Explanation

Correct Answer

A:

Recovery readiness depends on known-good artifacts and practiced restoration steps; production testing should demonstrate that a drill restores the preceding prompt-model-index combination without inconsistent state.

Incorrect Answers

B:

Retrieval behavior cannot be recreated from the generation prompt alone; this makes sense when a team documents prompts but not which corpus, embedding model and chunking policy produced answers, while this problem involves a service outage requires rollback, but deployment notes contain no previous artifact versions.

C:

Architecture assertions need operational artifacts to demonstrate the control actually runs; this technique fits where a compliance auditor requests proof that a human approval gate is enforced in production, and not where a service outage requires rollback, but deployment notes contain no previous artifact versions.

D:

A supportable deployment requires responsibility and recovery documentation beyond diagrams; the trigger would be that the delivery team leaves after launch with no record of tool privileges or secret owners, rather than the risk that a service outage requires rollback, but deployment notes contain no previous artifact versions.

Question 442

A solution has dozens of architectural assumptions recorded in chat messages only. Which implementation decision is the strongest fit?

  1. As a candidate control, supply sequence diagrams and failure-mode tables covering ordering, retries and idempotency.
  2. In the initial solution, create an operational handoff covering IAM, integrations, secrets, owners and incident escalation.
  3. In the proposed design, document the control design and link it to immutable approval-event evidence.
  4. Under the reviewed approach, maintain a living ADR register with owners, triggers and supersession history.

Explanation

Correct Answer

D:

Decision provenance reduces rediscovery when regulations, usage and models evolve; release assurance requires proof that engineers can find current and obsolete decisions with their reasons and review dates.

Incorrect Answers

A:

Multi-stage side effects need explicit operational contracts to prevent replay and duplication errors; that is defensible if an orchestrated workflow uses multiple asynchronous queues whose retry semantics are undocumented, whereas the current issue is that a solution has dozens of architectural assumptions recorded in chat messages only.

B:

A supportable deployment requires responsibility and recovery documentation beyond diagrams; the other workload is one where the delivery team leaves after launch with no record of tool privileges or secret owners, whereas the release criterion addresses a solution has dozens of architectural assumptions recorded in chat messages only.

C:

Architecture assertions need operational artifacts to demonstrate the control actually runs; this applies when a compliance auditor requests proof that a human approval gate is enforced in production, while the required safeguard concerns a solution has dozens of architectural assumptions recorded in chat messages only.

Question 443

A client receives final source code but no explanation of how to measure quality after deployment. What should be prioritized during architecture review?

  1. In the proposed implementation, supply sequence diagrams and failure-mode tables covering ordering, retries and idempotency.
  2. At this decision point, include rollback procedures, dependencies and version compatibility in the runbook.
  3. As a candidate control, handoff evaluation datasets, thresholds, dashboards and incident procedures with ownership.
  4. In the initial solution, version the RAG architecture and source lifecycle with reproducible index-build records.

Explanation

Correct Answer

C:

Operations must be able to assess ongoing correctness independently of the original builders; the intended outcome is that the receiving team executes a regression test and interprets the results without vendor help.

Incorrect Answers

A:

Multi-stage side effects need explicit operational contracts to prevent replay and duplication errors; it is relevant if an orchestrated workflow uses multiple asynchronous queues whose retry semantics are undocumented, instead of when a client receives final source code but no explanation of how to measure quality after deployment.

B:

Recovery readiness depends on known-good artifacts and practiced restoration steps; it serves conditions where a service outage requires rollback, but deployment notes contain no previous artifact versions, not the case in which a client receives final source code but no explanation of how to measure quality after deployment.

D:

Retrieval behavior cannot be recreated from the generation prompt alone; this makes sense when a team documents prompts but not which corpus, embedding model and chunking policy produced answers, but this decision concerns a client receives final source code but no explanation of how to measure quality after deployment.

Question 444

An orchestrated workflow uses multiple asynchronous queues whose retry semantics are undocumented. Which response follows from the actual operational constraint?

  1. In the release design, maintain a living ADR register with owners, triggers and supersession history.
  2. During architecture review, include rollback procedures, dependencies and version compatibility in the runbook.
  3. As the design priority, supply sequence diagrams and failure-mode tables covering ordering, retries and idempotency.
  4. At this decision point, handoff evaluation datasets, thresholds, dashboards and incident procedures with ownership.

Explanation

Correct Answer

C:

Multi-stage side effects need explicit operational contracts to prevent replay and duplication errors; its effect can be measured when an operations drill identifies which queue can retry safely after a partial failure.

Incorrect Answers

A:

Decision provenance reduces rediscovery when regulations, usage and models evolve; that would be useful if a solution has dozens of architectural assumptions recorded in chat messages only, while this problem involves an orchestrated workflow uses multiple asynchronous queues whose retry semantics are undocumented.

B:

Recovery readiness depends on known-good artifacts and practiced restoration steps; it serves a setting where a service outage requires rollback, but deployment notes contain no previous artifact versions, and not where an orchestrated workflow uses multiple asynchronous queues whose retry semantics are undocumented.

D:

Operations must be able to assess ongoing correctness independently of the original builders; the other workload is one where a client receives final source code but no explanation of how to measure quality after deployment, rather than the condition that an orchestrated workflow uses multiple asynchronous queues whose retry semantics are undocumented.

Question 445

An incident responder cannot locate who owns a production tool token or which team can rotate it. Which acceptance check evaluates this requirement rather than a proxy?

  1. A production-like test shows that a drill restores the preceding prompt-model-index combination without inconsistent state.
  2. The release check confirms that on-call staff can identify permission owners and rotate credentials using the runbook.
  3. The reviewer observes that a sampled transaction trace shows the required authorized approval before commit.
  4. The release report verifies that engineers can find current and obsolete decisions with their reasons and review dates.

Explanation

Correct Answer

B:

A supportable deployment requires responsibility and recovery documentation beyond diagrams; this is a functional check for the case where the delivery team leaves after launch with no record of tool privileges or secret owners.

Incorrect Answers

A:

Recovery readiness depends on known-good artifacts and practiced restoration steps; the approach helps when a service outage requires rollback, but deployment notes contain no previous artifact versions, whereas the release criterion addresses an incident responder cannot locate who owns a production tool token or which team can rotate it.

C:

Architecture assertions need operational artifacts to demonstrate the control actually runs; it is relevant if a compliance auditor requests proof that a human approval gate is enforced in production, not the risk where an incident responder cannot locate who owns a production tool token or which team can rotate it.

D:

Decision provenance reduces rediscovery when regulations, usage and models evolve; it serves conditions where a solution has dozens of architectural assumptions recorded in chat messages only, not this situation where an incident responder cannot locate who owns a production tool token or which team can rotate it.

Question 446

A knowledge-retrieval system is rebuilt from source control but returns different passages from the previous release. Which measurement is the most relevant release criterion?

  1. The observed outcome is that the receiving team executes a regression test and interprets the results without vendor help.
  2. The regression suite verifies that on-call staff can identify permission owners and rotate credentials using the runbook.
  3. A production-like test shows that a fresh environment rebuilds the index and reproduces known evidence retrieval.
  4. The release check confirms that an operations drill identifies which queue can retry safely after a partial failure.

Explanation

Correct Answer

C:

Retrieval behavior cannot be recreated from the generation prompt alone; the control should hold even if a team documents prompts but not which corpus, embedding model and chunking policy produced answers.

Incorrect Answers

A:

Operations must be able to assess ongoing correctness independently of the original builders; that fits cases where a client receives final source code but no explanation of how to measure quality after deployment, not the case in which a knowledge-retrieval system is rebuilt from source control but returns different passages from the previous release.

B:

A supportable deployment requires responsibility and recovery documentation beyond diagrams; this control is needed when the delivery team leaves after launch with no record of tool privileges or secret owners, whereas the problem presented is that a knowledge-retrieval system is rebuilt from source control but returns different passages from the previous release.

D:

Multi-stage side effects need explicit operational contracts to prevent replay and duplication errors; it serves a setting where an orchestrated workflow uses multiple asynchronous queues whose retry semantics are undocumented, but the actual risk is that a knowledge-retrieval system is rebuilt from source control but returns different passages from the previous release.

Question 447

An auditor asks for transaction-level proof that an authorized person approved a sensitive write. Which production-like test finding is most convincing?

  1. The scenario test demonstrates that a sampled transaction trace shows the required authorized approval before commit.
  2. The end-to-end check confirms that on-call staff can identify permission owners and rotate credentials using the runbook.
  3. The observed outcome is that a fresh environment rebuilds the index and reproduces known evidence retrieval.
  4. The regression suite verifies that the receiving team executes a regression test and interprets the results without vendor help.
  5. A production-like test shows that engineers can find current and obsolete decisions with their reasons and review dates.

Explanation

Correct Answer

A:

Architecture assertions need operational artifacts to demonstrate the control actually runs; it is the decisive test when a compliance auditor requests proof that a human approval gate is enforced in production.

Incorrect Answers

B:

A supportable deployment requires responsibility and recovery documentation beyond diagrams; it targets cases where the delivery team leaves after launch with no record of tool privileges or secret owners, rather than the risk that an auditor asks for transaction-level proof that an authorized person approved a sensitive write.

C:

Retrieval behavior cannot be recreated from the generation prompt alone; the approach helps when a team documents prompts but not which corpus, embedding model and chunking policy produced answers, rather than the condition that an auditor asks for transaction-level proof that an authorized person approved a sensitive write.

D:

Operations must be able to assess ongoing correctness independently of the original builders; that belongs to situations where a client receives final source code but no explanation of how to measure quality after deployment, while the immediate question involves an auditor asks for transaction-level proof that an authorized person approved a sensitive write.

E:

Decision provenance reduces rediscovery when regulations, usage and models evolve; it is relevant if a solution has dozens of architectural assumptions recorded in chat messages only, but the present constraint is that an auditor asks for transaction-level proof that an authorized person approved a sensitive write.

Question 448

A distributed retrieval workflow fails after an external write and is rolled back while queues still contain retries. Operators have not linked model, prompt and index versions, nor proved safe retry behavior. Which TWO drill results would establish reliable rollback and recovery? Choose TWO.

  1. A controlled trial finds that engineers can find current and obsolete decisions with their reasons and review dates.
  2. The measured evidence confirms that a fresh environment rebuilds the index and reproduces known evidence retrieval.
  3. The scenario test demonstrates that the receiving team executes a regression test and interprets the results without vendor help.
  4. The end-to-end check confirms that a drill restores the preceding prompt-model-index combination without inconsistent state.
  5. The observed outcome is that an operations drill identifies which queue can retry safely after a partial failure.

Explanation

Correct Answers

D:

Recovery readiness depends on known-good artifacts and practiced restoration steps; it demonstrates the required control when a service outage requires rollback, but deployment notes contain no previous artifact versions; another obligation arises when an orchestrated workflow uses multiple asynchronous queues whose retry semantics are undocumented.

E:

Multi-stage side effects need explicit operational contracts to prevent replay and duplication errors; the test covers the original risk that an orchestrated workflow uses multiple asynchronous queues whose retry semantics are undocumented; another obligation arises when a service outage requires rollback, but deployment notes contain no previous artifact versions.

Incorrect Answers

A:

Decision provenance reduces rediscovery when regulations, usage and models evolve; the trigger would be that a solution has dozens of architectural assumptions recorded in chat messages only, while the required safeguard concerns a rollback drill fails because the previous prompt, model and retrieval index are no longer associated; another obligation arises when an orchestrated workflow uses multiple asynchronous queues whose retry semantics are undocumented.

B:

Retrieval behavior cannot be recreated from the generation prompt alone; the design helps when a team documents prompts but not which corpus, embedding model and chunking policy produced answers, not the risk where a replay of an asynchronous workflow creates duplicate actions because no one recorded queue retry and idempotency semantics; another obligation arises when a service outage requires rollback, but deployment notes contain no previous artifact versions.

C:

Operations must be able to assess ongoing correctness independently of the original builders; that fits cases where a client receives final source code but no explanation of how to measure quality after deployment, not this situation where a rollback drill fails because the previous prompt, model and retrieval index are no longer associated; another obligation arises when an orchestrated workflow uses multiple asynchronous queues whose retry semantics are undocumented.

Question 449

A new maintainer inherits conflicting design notes and cannot tell which decision superseded an older one. Which verified outcome most directly addresses the reported risk?

  1. The representative case test finds that a fresh environment rebuilds the index and reproduces known evidence retrieval.
  2. The negative-case evaluation confirms that the receiving team executes a regression test and interprets the results without vendor help.
  3. The audit sample demonstrates that engineers can find current and obsolete decisions with their reasons and review dates.
  4. The measured evidence confirms that an operations drill identifies which queue can retry safely after a partial failure.

Explanation

Correct Answer

C:

Decision provenance reduces rediscovery when regulations, usage and models evolve; the observed protection matters when a solution has dozens of architectural assumptions recorded in chat messages only.

Incorrect Answers

A:

Retrieval behavior cannot be recreated from the generation prompt alone; the alternative helps if a team documents prompts but not which corpus, embedding model and chunking policy produced answers, whereas the problem presented is that a new maintainer inherits conflicting design notes and cannot tell which decision superseded an older one.

B:

Operations must be able to assess ongoing correctness independently of the original builders; it is justified if a client receives final source code but no explanation of how to measure quality after deployment, but this decision concerns a new maintainer inherits conflicting design notes and cannot tell which decision superseded an older one.

D:

Multi-stage side effects need explicit operational contracts to prevent replay and duplication errors; it targets cases where an orchestrated workflow uses multiple asynchronous queues whose retry semantics are undocumented, but the requested boundary concerns a new maintainer inherits conflicting design notes and cannot tell which decision superseded an older one.

Question 450

A customer maintenance team receives application code but cannot run the model quality checks on its own. What independently verifiable observation is needed?

  1. The verified result is that engineers can find current and obsolete decisions with their reasons and review dates.
  2. The recorded test establishes that a fresh environment rebuilds the index and reproduces known evidence retrieval.
  3. The representative case test finds that a drill restores the preceding prompt-model-index combination without inconsistent state.
  4. The staged rollout demonstrates that the receiving team executes a regression test and interprets the results without vendor help.

Explanation

Correct Answer

D:

Operations must be able to assess ongoing correctness independently of the original builders; the safeguard must work when a client receives final source code but no explanation of how to measure quality after deployment.

Incorrect Answers

A:

Decision provenance reduces rediscovery when regulations, usage and models evolve; this makes sense when a solution has dozens of architectural assumptions recorded in chat messages only, rather than the risk that a customer maintenance team receives application code but cannot run the model quality checks on its own.

B:

Retrieval behavior cannot be recreated from the generation prompt alone; this technique fits where a team documents prompts but not which corpus, embedding model and chunking policy produced answers, rather than the condition that a customer maintenance team receives application code but cannot run the model quality checks on its own.

C:

Recovery readiness depends on known-good artifacts and practiced restoration steps; the trigger would be that a service outage requires rollback, but deployment notes contain no previous artifact versions, while the immediate question involves a customer maintenance team receives application code but cannot run the model quality checks on its own.

Claude Certified Architect - Professional (CCAR-P) Practice Tests