Anthropic CCAR-P (Claude Certified Architect - Professional) Practice Test 36

Question 350

A financial research assistant now performs two independent evidence lookups sequentially even though neither depends on the other. What performance consequence comes from dropping concurrent retrieval?

  1. A failure where a fixed high budget can waste time and tokens on simple cases
  2. A failure where holding an interactive connection open invites timeouts and duplicate retries
  3. A failure where cutting retrieval silently may produce unsupported definitive statements
  4. A failure where sequential calls inflate end-to-end wait time without improving correctness

Explanation

Correct Answer

D:

The relevant failure mode is that sequential calls inflate end-to-end wait time without improving correctness. The existing design can prevent it when the integration can fetch independent sources concurrently and join their authorized results. The team needs independent retrievers and bounded parallelism and must monitor critical-path retrieval latency and source coverage to support any claim of equivalent protection.

Incorrect Answers

A:

If straightforward lookup requests and deep analysis use very different effort, a fixed high budget can waste time and tokens on simple cases; but this is not the immediate loss associated with omitting Parallel independent retrieval: sequential calls inflate end-to-end wait time without improving correctness.

B:

When complex report synthesis exceeds an interactive HTTP timeout, holding an interactive connection open invites timeouts and duplicate retries; however, the named safeguard addresses the distinct hazard that sequential calls inflate end-to-end wait time without improving correctness.

C:

When a strict response-time target sometimes conflicts with deep evidence searches, cutting retrieval silently may produce unsupported definitive statements; whereas removing Parallel independent retrieval primarily exposes the risk that sequential calls inflate end-to-end wait time without improving correctness.

Question 351

For a research administration production change, the required effect is that structured error tags make future fixes attributable to retrieval, prompt or policy causes. Which TWO selections jointly identify the intervention and its enabling prerequisite? Choose TWO.

  1. Implement Feedback taxonomy with ground-truth capture
  2. Establish trained reviewers and a stable reason-code dictionary
  3. Track identifier leakage rate on representative test records
  4. Implement Typed output validation and repair route
  5. Establish an explicit schema and ownership of input validation

Explanation

Correct Answers

A:

This is the design control: structured error tags make future fixes attributable to retrieval, prompt or policy causes. It addresses the hazard that undifferentiated feedback cannot guide targeted changes; the other selected element supplies the deployment foundation and, where required, the measurement of that control.

B:

This step creates the enabling basis of trained reviewers and a stable reason-code dictionary. That preparation supports the mechanism that structured error tags make future fixes attributable to retrieval, prompt or policy causes; merely choosing the control does not prove its configuration or authority is ready.

Incorrect Answers

C:

This test would validate a different benefit: preprocessing redacts unnecessary fields while preserving task-bearing content; but that indicator does not verify share of reviewed errors with actionable classification for the selected architecture.

D:

This design choice is aimed at cases where downstream software requires machine-readable dispositions with fixed enums; it is effective because the architecture can validate generated payloads against a contract and route failures to repair or review. The rollout described is instead governed by the case where reviewers only click thumbs-up or thumbs-down without reason codes and requires a different primary intervention; however, the intended intervention requires that structured error tags make future fixes attributable to retrieval, prompt or policy causes.

E:

The alternative setup enables a different mechanism: schema checks and normalization reject malformed inputs before they enter the model context; whereas this intervention requires trained reviewers and a stable reason-code dictionary as its operating precondition.

Question 352

The field-service coordination go-live board accepts the premise that the model configuration can enforce authorization and policy in deterministic application controls. Which preparation is indispensable before it can rely on Guardrail outside generation operationally?

  1. Establish server-side policy enforcement before side effects
  2. Establish vetted examples covering ambiguous categories
  3. Establish immutable prompt registry and release identifiers
  4. Establish a deterministic postprocessor and regional format rules

Explanation

Correct Answer

A:

Readiness depends on server-side policy enforcement before side effects, not just a design approval. That condition supports the mechanism in which the model configuration can enforce authorization and policy in deterministic application controls. Without it, a warning in the prompt does not prevent a tool executor from making a prohibited change; monitoring prevented unauthorized actions during adversarial tests helps verify the deployed result.

Incorrect Answers

B:

This prerequisite supports a method used when a document tagger confuses closely related exception labels, because the model configuration can demonstrate boundary cases with labeled examples near the instruction; but the production dependency for Guardrail outside generation is server-side policy enforcement before side effects.

C:

This prerequisite supports a method used when deployments silently change instruction wording between evaluation and production, because the model configuration can pin and trace prompt versions through release and rollback; for this release, the team must first verify server-side policy enforcement before side effects.

D:

This prerequisite supports a method used when downstream clients need stable currency and date conventions, because the model configuration can normalize and validate final fields after model generation; however, the named intervention needs server-side policy enforcement before side effects before its protection can be enforced.

Question 353

During solution discovery in field-service coordination, policy retrieval quality looks good using documents from the test period. Which proposed approach handles this situation at its source?

  1. Held-out evaluation split
  2. Stratified production-like test set
  3. Adversarial injection fixtures
  4. Temporal holdout for changing policies

Explanation

Correct Answer

D:

The intervention acts on the system behavior directly: the evaluation process can test on later source changes to measure freshness and drift response. Before rollout, verify dated corpora and known change events. The relevant success test is version-correct answers after policy refresh; otherwise static snapshots fail to reveal stale-index behavior.

Incorrect Answers

A:

When prompt engineers tune repeatedly on the same example corpus, the evaluation process can reserve untouched cases for final comparison after iteration; whereas the field-service coordination case must guard against the outcome that static snapshots fail to reveal stale-index behavior.

B:

When long-tail failure classes rarely appear in a random convenience sample, the evaluation process can stratify by task type, risk, language and source complexity; the alternative leaves the central requirement unproven: the evaluation process can test on later source changes to measure freshness and drift response.

C:

If a tool agent ingests external documents that can contain hostile instructions, the evaluation process can add attacker-controlled content and escalation attempts to evaluation; it does not ensure the required behavior: the evaluation process can test on later source changes to measure freshness and drift response.

Question 354

A software platform service is moving to production with this aim: the delivery team can explain residual risk, enforceable mitigations and explicit escalation boundaries. Choose TWO complementary choices, an architectural control and the precondition it depends on. Choose TWO.

  1. Establish common risk taxonomy and mitigation evidence
  2. Implement Risk-and-mitigation narrative
  3. Establish representative scenarios and evaluation criteria
  4. Track decisions traced to approved evidence and reversal triggers
  5. Implement TCO and change-cost briefing

Explanation

Correct Answers

A:

Choosing this setup item establishes common risk taxonomy and mitigation evidence, which is needed for reliable operation when the delivery team can explain residual risk, enforceable mitigations and explicit escalation boundaries. The Risk-and-mitigation narrative rollout should not be declared ready until common risk taxonomy and mitigation evidence is verified in its operating environment.

B:

This is the design control: the delivery team can explain residual risk, enforceable mitigations and explicit escalation boundaries. It addresses the hazard that generic assurances about safety do not settle operational accountability; the other selected element supplies the deployment foundation and, where required, the measurement of that control.

Incorrect Answers

C:

The alternative setup enables a different mechanism: the delivery team can compare alternatives using concrete failure scenarios rather than vague preferences; whereas this intervention requires common risk taxonomy and mitigation evidence as its operating precondition.

D:

This test would validate a different benefit: the delivery team can record context, alternatives, decision and consequences in a versioned ADR; but that indicator does not verify signed risk acceptance and validated control ownership for the selected architecture.

E:

This design choice is aimed at cases where a model switch looks cheap but requires extensive template and integration rework; it is effective because the delivery team can include maintenance, migration, vendor dependencies and personnel effort in comparison; however, the intended intervention requires that the delivery team can explain residual risk, enforceable mitigations and explicit escalation boundaries.

Question 355

A software platform audit challenges the need for Data exposure and retention review. Which risk would remain unmanaged if its function were not provided elsewhere?

  1. The risk that persuasive wording can mislead users into assuming unwarranted certainty
  2. The risk that encryption in transit alone does not resolve overcollection or retention
  3. The risk that a one-time approval cannot guarantee future behavior across drift
  4. The risk that a human in the workflow may not provide meaningful oversight

Explanation

Correct Answer

B:

The relevant failure mode is that encryption in transit alone does not resolve overcollection or retention. The existing design can prevent it when the control layer can trace data paths, access, retention and deletion obligations before launch. The team needs approved data-flow inventory and service agreements and must monitor unauthorized data persistence and access findings to support any claim of equivalent protection.

Incorrect Answers

A:

If an assistant states confident conclusions without revealing evidence limitations, persuasive wording can mislead users into assuming unwarranted certainty; whereas removing Data exposure and retention review primarily exposes the risk that encryption in transit alone does not resolve overcollection or retention.

C:

If new prompt versions or changing input mix alter policy compliance, a one-time approval cannot guarantee future behavior across drift; however, the named safeguard addresses the distinct hazard that encryption in transit alone does not resolve overcollection or retention.

D:

When reviewers rubber-stamp model recommendations under time pressure, a human in the workflow may not provide meaningful oversight; whereas removing Data exposure and retention review primarily exposes the risk that encryption in transit alone does not resolve overcollection or retention.

Question 356

In a procurement services modernization project, complex report synthesis exceeds an interactive HTTP timeout. Which approach best fits this specific operating constraint?

  1. Deploy Adaptive reasoning budget
  2. Deploy Parallel independent retrieval
  3. Deploy Latency-aware fallback answer
  4. Deploy Rerank only the candidate set
  5. Deploy Selective verification for risky outputs
  6. Deploy Asynchronous completion for long tasks

Explanation

Correct Answer

F:

The mechanism is specific: the integration can submit a job and return an identifier with status tracking. A credible implementation requires persistent job state and idempotent submission, while its acceptance test monitors completion success and queue wait under load. The alternative approaches leave the risk that holding an interactive connection open invites timeouts and duplicate retries.

Incorrect Answers

A:

When straightforward lookup requests and deep analysis use very different effort, the integration can allocate larger reasoning budgets only to tasks that need them; whereas the procurement services case must guard against the outcome that holding an interactive connection open invites timeouts and duplicate retries.

B:

When an answer needs evidence from several unrelated repositories, the integration can fetch independent sources concurrently and join their authorized results; the missing operational behavior is specific: the integration can submit a job and return an identifier with status tracking.

C:

For a workload where a strict response-time target sometimes conflicts with deep evidence searches, the integration can provide an explicit partial or queued result rather than fabricate a quick full answer; the current decision instead depends on a specific capability: the integration can submit a job and return an identifier with status tracking.

D:

For a workload where broad retrieval finds relevant content but top results contain noisy matches, the integration can rerank a bounded candidate list for precision before generation; whereas the procurement services case must guard against the outcome that holding an interactive connection open invites timeouts and duplicate retries.

E:

When routine descriptions are low stakes but payment changes require evidence checking, the integration can apply expensive verification only to high-impact or low-confidence actions; the missing operational behavior is specific: the integration can submit a job and return an identifier with status tracking.

Question 357

In a field-service coordination modernization project, a retrying agent sometimes creates the same support task twice. Which approach best fits this specific operating constraint?

  1. Idempotent action boundary
  2. Preprocessing and sensitive-field minimization
  3. Explicit intermediate state store
  4. Audit-correlated processing stages

Explanation

Correct Answer

A:

This control changes the relevant system boundary: idempotency keys and deduplication protect side-effecting writes across retries. The team should establish stable request identifiers and a deduplication store before enabling it, then evaluate duplicate external writes per retried request so that a network timeout can trigger repeated external state changes is not hidden by fluent outputs.

Incorrect Answers

B:

For a workload where raw customer messages contain identifiers irrelevant to classification, preprocessing redacts unnecessary fields while preserving task-bearing content; the stated requirement remains unmet: idempotency keys and deduplication protect side-effecting writes across retries.

C:

When multi-step reviews lose progress during retries and restarts, durable state records completed stages and enables safe continuation; whereas the field-service coordination case must guard against the outcome that a network timeout can trigger repeated external state changes.

D:

If compliance teams cannot reconstruct which tool outputs informed a recommendation, stage-level correlation links input versions, retrieval results, model calls and final decisions; this rollout still needs the following guarantee: idempotency keys and deduplication protect side-effecting writes across retries.

Question 358

The education services deployment uses Guardrail outside generation, but executive reports currently count only successful model responses. The residual concern is that a warning in the prompt does not prevent a tool executor from making a prohibited change. Which acceptance measure would reveal real progress?

  1. Evaluate output consistency across template versions and inputs
  2. Evaluate schema validity and calibrated abstention on impossible cases
  3. Evaluate prevented unauthorized actions during adversarial tests
  4. Evaluate macro F1 on confusable label pairs

Explanation

Correct Answer

C:

The release should evaluate prevented unauthorized actions during adversarial tests rather than equate successful calls with real outcomes. The control works on the premise that the model configuration can enforce authorization and policy in deterministic application controls; the proposed measure can expose the risk that a warning in the prompt does not prevent a tool executor from making a prohibited change.

Incorrect Answers

A:

The proposed measure would suit a setting where hundreds of forms share the same logic with varying account details, since the model configuration can keep stable instructions versioned and inject only validated task variables; but the acceptance gate must instead inspect prevented unauthorized actions during adversarial tests to detect the targeted failure.

B:

The alternative indicator checks whether the model configuration can define output fields and abstention behavior for unanswerable inputs when the caller needs a machine-readable response even when evidence is insufficient; by contrast, the education services release needs evidence from prevented unauthorized actions during adversarial tests.

D:

The proposed measure would suit a setting where a document tagger confuses closely related exception labels, since the model configuration can demonstrate boundary cases with labeled examples near the instruction; but the acceptance gate must instead inspect prevented unauthorized actions during adversarial tests to detect the targeted failure.

Question 359

The public-sector casework go-live board accepts the premise that the evaluation process can test on later source changes to measure freshness and drift response. Which preparation is indispensable before it can rely on Temporal holdout for changing policies operationally?

  1. Establish leakage-resistant dataset versioning
  2. Establish dated corpora and known change events
  3. Establish privacy-cleared examples and a sampling design
  4. Establish an explicit list of semantics-preserving changes
  5. Establish frozen consumer contract and test harness

Explanation

Correct Answer

B:

The required preparation is dated corpora and known change events. It makes the following architecture claim enforceable: the evaluation process can test on later source changes to measure freshness and drift response. If that setup is skipped, static snapshots fail to reveal stale-index behavior, so the release should examine version-correct answers after policy refresh.

Incorrect Answers

A:

This prerequisite supports a method used when prompt engineers tune repeatedly on the same example corpus, because the evaluation process can reserve untouched cases for final comparison after iteration; but the production dependency for Temporal holdout for changing policies is dated corpora and known change events.

C:

For the separate challenge where long-tail failure classes rarely appear in a random convenience sample, the evaluation process can stratify by task type, risk, language and source complexity; however, the named intervention needs dated corpora and known change events before its protection can be enforced.

D:

This prerequisite supports a method used when answers should not change when irrelevant formatting or names change, because the evaluation process can apply controlled input transformations and compare invariant outcomes; but the production dependency for Temporal holdout for changing policies is dated corpora and known change events.

E:

The setup is useful to enable the alternative's behavior where the evaluation process can validate outputs mechanically on every release candidate and downstream software expects exact structured fields and allowed enum values; for this release, the team must first verify dated corpora and known change events.

Question 360

A software platform team is advancing Asynchronous completion for long tasks beyond its proof of concept. Engineers expect that the integration can submit a job and return an identifier with status tracking. What evidence of setup readiness should be required first?

  1. Confirm explicit risk classifier and verification budget
  2. Confirm persistent job state and idempotent submission
  3. Confirm independent retrievers and bounded parallelism
  4. Confirm quality-calibrated routing and budget caps

Explanation

Correct Answer

B:

The prerequisite creates the necessary operating boundary: persistent job state and idempotent submission. The system is intended to work because the integration can submit a job and return an identifier with status tracking. A release without that foundation may still suffer holding an interactive connection open invites timeouts and duplicate retries; use completion success and queue wait under load to evaluate the result.

Incorrect Answers

A:

The setup is useful to enable the alternative's behavior where the integration can apply expensive verification only to high-impact or low-confidence actions and routine descriptions are low stakes but payment changes require evidence checking; for this release, the team must first verify persistent job state and idempotent submission.

C:

Where an answer needs evidence from several unrelated repositories, the integration can fetch independent sources concurrently and join their authorized results; but the production dependency for Asynchronous completion for long tasks is persistent job state and idempotent submission.

D:

Where straightforward lookup requests and deep analysis use very different effort, the integration can allocate larger reasoning budgets only to tasks that need them; for this release, the team must first verify persistent job state and idempotent submission.

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