Anthropic CCAR-P (Claude Certified Architect - Professional) Practice Test 5

Question 40

The defining problem for a compliance assurance pilot is that assistants propose privileged write operations in customer systems. Which proposed approach addresses it most precisely?

  1. Introduce Untrusted retrieved-content isolation
  2. Introduce High-risk human approval gate
  3. Introduce Tool-execution policy gate
  4. Introduce Output privacy filter
  5. Introduce Defense-in-depth incident response

Explanation

Correct Answer

C:

The intervention acts on the system behavior directly: the control layer can require server-side authorization and business-policy checks before each side effect. Before rollout, verify central policy service and auditable subject identity. The relevant success test is disallowed action attempts stopped at the executor; otherwise model self-restraint cannot guarantee that a bad action is blocked.

Incorrect Answers

A:

For a workload where supplier documents include embedded instructions masquerading as system messages, the control layer can treat retrieved text as evidence and prohibit it from changing tool policy; the alternative cannot replace this key requirement: the control layer can require server-side authorization and business-policy checks before each side effect.

B:

For a workload where proposed external transfers could produce irreversible customer harm, the control layer can stop before execution and require authorized human confirmation; whereas the compliance assurance case must guard against the outcome that model self-restraint cannot guarantee that a bad action is blocked.

D:

When generated summaries may echo sensitive identifiers unnecessary for recipients, the control layer can enforce recipient-specific redaction and validated output fields; the alternative cannot replace this key requirement: the control layer can require server-side authorization and business-policy checks before each side effect.

E:

When failures may bypass one input or model safety layer, the control layer can combine prevention, detection, containment and audited recovery; whereas the compliance assurance case must guard against the outcome that model self-restraint cannot guarantee that a bad action is blocked.

Question 41

A field-service coordination release candidate fails review after assessors observe that eager inclusion of the entire catalog wastes context and confuses selection. What intervention should take priority?

  1. Enforce Progressive tool discovery
  2. Enforce Task-scoped tool allowlist
  3. Enforce Read-write capability separation
  4. Enforce Single-purpose tool contract

Explanation

Correct Answer

A:

The engineering fix changes the mechanism rather than the presentation: the integration can advertise compact discovery metadata and load detailed definitions on demand. It requires curated discoverability metadata and scoped authorization; a regression test should examine tool-definition tokens per resolved user task in cases where eager inclusion of the entire catalog wastes context and confuses selection.

Incorrect Answers

B:

The alternative reduces the risk that broad tool visibility increases selection error and attack surface because the integration can expose only tools needed for the active task and user permission scope; however, the field-service coordination defect follows the path in which eager inclusion of the entire catalog wastes context and confuses selection.

C:

When mixing reading and mutation in one tool overgrants routine sessions, the integration can expose read-only access independently from privileged changes; yet it would not correct the observed weakness: eager inclusion of the entire catalog wastes context and confuses selection.

D:

This addresses the failure where opaque universal tools hide security boundaries and failure semantics, because the integration can replace broad verbs with narrow parameterized operations and validation; but this remedy operates on another boundary, not the case where eager inclusion of the entire catalog wastes context and confuses selection.

Question 42

A telecommunications support pilot has enabled Grounded-claim verification rate. Because citation formatting alone does not prove support for the claim, the release board wants evidence of effectiveness, not just a favorable demonstration. Which metric most directly tests the proposed control?

  1. Evaluate high-percentile latency and timeout frequency
  2. Evaluate verified factual claims divided by audited claims
  3. Evaluate risk-coverage curve and improper certainty rate
  4. Evaluate false-allow and false-deny rates by scope

Explanation

Correct Answer

B:

Tracking verified factual claims divided by audited claims provides direct evidence that the system delivers its intended effect: the evaluation process can score whether factual claims are supported by cited source spans. The measure is more informative than prompt counts because the failure being measured is that citation formatting alone does not prove support for the claim.

Incorrect Answers

A:

The alternative indicator checks whether the evaluation process can track p95 and p99 end-to-end latency by request class when interactive sessions frequently exceed their response deadline; however, that proxy would not demonstrate verified factual claims divided by audited claims for the selected intervention.

C:

Where incomplete evidence should yield escalation rather than invented certainty, a relevant check would establish whether the evaluation process can evaluate correctness jointly with justified abstentions across uncertainty levels; by contrast, the telecommunications support release needs evidence from verified factual claims divided by audited claims.

D:

The proposed measure would suit a setting where agent integration must respect access restrictions for multiple tenants, since the evaluation process can evaluate denied and erroneously permitted requests against authoritative policy; however, that proxy would not demonstrate verified factual claims divided by audited claims for the selected intervention.

Question 43

The defining problem for a legal operations pilot is that multi-document arbitration requires several interdependent inferences. Which proposed approach addresses it most precisely?

  1. Deploy Small-model first-pass routing
  2. Deploy Context-capacity fit before model selection
  3. Deploy Per-class model policy
  4. Deploy Reasoning-capable model for complex synthesis
  5. Deploy Empirical model bake-off

Explanation

Correct Answer

D:

The mechanism is specific: a stronger model can justify tradeoffs across heterogeneous evidence. A credible implementation requires benchmark cases with multi-hop evidence, while its acceptance test monitors audited synthesis accuracy on difficult cases. The alternative approaches leave the risk that a lightweight classifier can omit crucial cross-document reasoning.

Incorrect Answers

A:

When most messages are short and low-complexity while a minority need deep analysis, a fast adequate tier handles easy cases and escalates only uncertain ones; it does not ensure the required behavior: a stronger model can justify tradeoffs across heterogeneous evidence.

B:

For a workload where legal analysis requests require unusually long source packets, the model configuration can test effective context utilization and recall before choosing a model configuration; whereas the legal operations case must guard against the outcome that a lightweight classifier can omit crucial cross-document reasoning.

C:

When privacy-sensitive and low-risk workloads require different operating controls, the model configuration can set explicit allowed model configurations by data class and purpose; the alternative leaves the central requirement unproven: a stronger model can justify tradeoffs across heterogeneous evidence.

E:

For a workload where stakeholders choose a model based on brand rather than target workload, the model configuration can evaluate candidate models on identical task data and operating constraints; whereas the legal operations case must guard against the outcome that a lightweight classifier can omit crucial cross-document reasoning.

Question 44

After installing Feasibility pilot with representative cases in a healthcare operations workflow, the team must validate whether it has reduced the exposure that a polished demo can hide unacceptable long-tail failures. Which measurement is most relevant?

  1. Measure reviewer override and acceptance rates by decision category
  2. Measure risk-adjusted impact and implementation effort per use case
  3. Measure proportion of prohibited workflows rejected before execution
  4. Measure pass rate on a stratified set of authentic exceptions

Explanation

Correct Answer

D:

The release should evaluate pass rate on a stratified set of authentic exceptions rather than equate successful calls with real outcomes. The control works on the premise that small-scope trials expose feasibility and exception handling before a broad commitment; the proposed measure can expose the risk that a polished demo can hide unacceptable long-tail failures.

Incorrect Answers

A:

The proposed measure would suit a setting where caseworkers must retain authority for consequential final decisions, since Claude drafts ranked recommendations while a designated reviewer authorizes each action; by contrast, the healthcare operations release needs evidence from pass rate on a stratified set of authentic exceptions.

B:

The proposed measure would suit a setting where candidate use cases differ in savings, uncertainty and personal-data exposure, since a scored decision framework surfaces high-value lower-risk starting points; however, that proxy would not demonstrate pass rate on a stratified set of authentic exceptions for the selected intervention.

C:

The alternative indicator checks whether identifying out-of-scope judgments prevents unsupported automation promises when a client demands a guarantee of fully accurate autonomous legal advice; but the acceptance gate must instead inspect pass rate on a stratified set of authentic exceptions to detect the targeted failure.

Question 45

A travel operations architecture gate expects the following behavior: the integration can advertise compact discovery metadata and load detailed definitions on demand. Which TWO actions together supply the design intervention and its operating dependency? Choose TWO.

  1. Implement Task-scoped tool allowlist
  2. Establish curated discoverability metadata and scoped authorization
  3. Track successful read tasks without elevated write privilege
  4. Implement Progressive tool discovery
  5. Establish loop policy and classified tool-error responses

Explanation

Correct Answers

B:

Choosing this setup item establishes curated discoverability metadata and scoped authorization, which is needed for reliable operation when the integration can advertise compact discovery metadata and load detailed definitions on demand. The Progressive tool discovery rollout should not be declared ready until curated discoverability metadata and scoped authorization is verified in its operating environment.

D:

Implementing the intervention changes the operational mechanism so that the integration can advertise compact discovery metadata and load detailed definitions on demand. The failure to prevent is that eager inclusion of the entire catalog wastes context and confuses selection; configuration and validation remain separate release obligations.

Incorrect Answers

A:

This design choice is aimed at cases where a research agent is offered dozens of irrelevant mutation tools; it is effective because the integration can expose only tools needed for the active task and user permission scope. The rollout described is instead governed by the case where; however, the intended intervention requires that the integration can advertise compact discovery metadata and load detailed definitions on demand.

C:

This test would validate a different benefit: the integration can expose read-only access independently from privileged changes; but that indicator does not verify tool-definition tokens per resolved user task for the selected architecture.

E:

The alternative setup enables a different mechanism: the integration can enforce retry, token and call budgets with a deliberate stop path; whereas this intervention requires curated discoverability metadata and scoped authorization as its operating precondition.

Question 46

For a logistics coordination rollout, the proposed design uses Stakeholder and decision-rights mapping, relying on the claim that the delivery team can define accountable decision owners and consultation rights early. What enabling control must be in place to make that design enforceable?

  1. Establish access to operators and representative case samples
  2. Establish cross-functional workflow mapping sessions
  3. Establish stakeholder register and governance charter
  4. Establish accessible data catalog and source stewards

Explanation

Correct Answer

C:

The prerequisite creates the necessary operating boundary: stakeholder register and governance charter. The system is intended to work because the delivery team can define accountable decision owners and consultation rights early. A release without that foundation may still suffer unclear authority causes late redesign and unowned risk decisions; use outstanding decision-owner ambiguities before design signoff to evaluate the result.

Incorrect Answers

A:

This prerequisite supports a method used when sponsors describe an ideal process that differs from daily operations, because the delivery team can observe actual exception handling and frontline work before mapping automation; however, the named intervention needs stakeholder register and governance charter before its protection can be enforced.

B:

For the separate challenge where automation is proposed without mapping present handoffs and failure paths, the delivery team can chart the existing input-to-outcome process and exception volumes; but the production dependency for Stakeholder and decision-rights mapping is stakeholder register and governance charter.

D:

This prerequisite supports a method used when discovery assumes reliable source records that are fragmented or stale, because the delivery team can validate quality, permissions, update cadence and ownership before promising RAG results; however, the named intervention needs stakeholder register and governance charter before its protection can be enforced.

Question 47

The solution design for software platform assumes that the control layer can require server-side authorization and business-policy checks before each side effect. Before treating Tool-execution policy gate as production-ready, which dependency needs an owner and verification?

  1. Prepare recipient entitlement map and redaction tests
  2. Prepare layered controls and assigned incident owners
  3. Prepare documented trust boundaries and security test cases
  4. Prepare central policy service and auditable subject identity
  5. Prepare reviewer authority, timeout path and approval evidence

Explanation

Correct Answer

D:

Before deployment the team must confirm central policy service and auditable subject identity. Without that preparation, the claim that the control layer can require server-side authorization and business-policy checks before each side effect may remain unenforceable; the related exposure is that model self-restraint cannot guarantee that a bad action is blocked. The acceptance indicator is disallowed action attempts stopped at the executor.

Incorrect Answers

A:

For the separate challenge where generated summaries may echo sensitive identifiers unnecessary for recipients, the control layer can enforce recipient-specific redaction and validated output fields; but the production dependency for Tool-execution policy gate is central policy service and auditable subject identity.

B:

For the separate challenge where failures may bypass one input or model safety layer, the control layer can combine prevention, detection, containment and audited recovery; for this release, the team must first verify central policy service and auditable subject identity.

C:

The setup is useful to enable the alternative's behavior where the control layer can classify untrusted inputs and prevent them from acquiring instruction authority and user-submitted attachments can contain instructions to steal private data; however, the named intervention needs central policy service and auditable subject identity before its protection can be enforced.

E:

Where proposed external transfers could produce irreversible customer harm, the control layer can stop before execution and require authorized human confirmation; for this release, the team must first verify central policy service and auditable subject identity.

Question 48

An education services release candidate fails review after assessors observe that citation formatting alone does not prove support for the claim. What intervention should take priority?

  1. Enforce Calibration and abstention quality
  2. Enforce Security policy violation rate
  3. Enforce Task-level acceptance rubric
  4. Enforce Grounded-claim verification rate
  5. Enforce Unsafe-action escape rate

Explanation

Correct Answer

D:

This is the causal correction: the evaluation process can score whether factual claims are supported by cited source spans. With approved enabling configuration and accountable owner in place, the team can test verified factual claims divided by audited claims and show that the problematic case in which citation formatting alone does not prove support for the claim has been addressed.

Incorrect Answers

A:

This addresses the failure where accuracy on answered cases alone can reward refusing everything, because the evaluation process can evaluate correctness jointly with justified abstentions across uncertainty levels; whereas the education services defect follows the path in which citation formatting alone does not prove support for the claim.

B:

Where agent integration must respect access restrictions for multiple tenants, the evaluation process can evaluate denied and erroneously permitted requests against authoritative policy; yet it would not correct the observed weakness: citation formatting alone does not prove support for the claim.

C:

Although the alternative reduces the risk that generic satisfaction scores may reward confident errors because the evaluation process can define observable criteria for a usable output rather than aesthetic fluency; but this remedy operates on another boundary, not the case where citation formatting alone does not prove support for the claim.

E:

This addresses the failure where general answer accuracy can hide rare severe tool abuse, because the evaluation process can measure disallowed actions that reach a protected side-effecting endpoint; yet it would not correct the observed weakness: citation formatting alone does not prove support for the claim.

Question 49

A legal operations incident exposes a specific weakness: personal chat instructions do not establish reproducible team conventions. Which architectural correction addresses the cause rather than masking the symptom?

  1. Deploy Workspace separation for sensitive tasks
  2. Deploy Least-privilege tool permissions
  3. Deploy Repository-level project instructions
  4. Deploy Reproducible development environment

Explanation

Correct Answer

C:

The problem lies in the absence of a mechanism where the engineering workflow can version shared Claude Code guidance alongside the project and review changes. Reinstating it, backed by approved project guidance and code review owners, limits the exposure that personal chat instructions do not establish reproducible team conventions; the engineering team should track standards adherence across fresh developer workspaces.

Incorrect Answers

A:

This addresses the failure where a shared default environment increases cross-project exposure, because the engineering workflow can isolate project roots, secrets and tool permissions by approved workspace; yet it would not correct the observed weakness: personal chat instructions do not establish reproducible team conventions.

B:

Where coding agents request shell access beyond the repository task, the engineering workflow can configure approved actions and deny risky operations by default; but this remedy operates on another boundary, not the case where personal chat instructions do not establish reproducible team conventions.

D:

When ambient local settings conceal incompatibilities in shared builds, the engineering workflow can pin dependencies, tooling and environment assumptions across contributors; yet it would not correct the observed weakness: personal chat instructions do not establish reproducible team conventions.

Question 50

A retail support team is selecting the next improvement for its AI service; a generic execute-any-query tool lets a model submit unreviewed commands. What design choice has the strongest direct justification?

  1. Implement Progressive tool discovery
  2. Implement Single-purpose tool contract
  3. Implement Tool ownership and lifecycle review
  4. Implement Task-scoped tool allowlist

Explanation

Correct Answer

B:

The proposed control fits because the integration can replace broad verbs with narrow parameterized operations and validation. Its production prerequisites include documented action schemas and server-side validation; check prohibited-call rejection and schema-compliant invocation rates on representative tasks. Without this boundary, opaque universal tools hide security boundaries and failure semantics.

Incorrect Answers

A:

For a workload where a complex connector ecosystem cannot fit full tool descriptions in each request, the integration can advertise compact discovery metadata and load detailed definitions on demand; whereas the software platform case must guard against the outcome that opaque universal tools hide security boundaries and failure semantics.

C:

If several plugins expose overlapping outdated customer lookup functions, the integration can retire or consolidate redundant capabilities with a responsible owner; the alternative cannot replace this key requirement: the integration can replace broad verbs with narrow parameterized operations and validation.

D:

If a research agent is offered dozens of irrelevant mutation tools, the integration can expose only tools needed for the active task and user permission scope; whereas the software platform case must guard against the outcome that opaque universal tools hide security boundaries and failure semantics.

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