Anthropic CCAR-P (Claude Certified Architect - Professional) Practice Test 76

Question 750

The telecommunications support deployment uses Parent-child document retrieval, but executive reports currently count only successful model responses. The residual concern is that returning only tiny isolated fragments removes governing context. Which acceptance measure would reveal real progress?

  1. Track full-section citation accuracy at a fixed context budget
  2. Track time-to-index-freshness and stale-answer incidence
  3. Measure precision of retrieval after authorization filtering
  4. Track recall on exact-code and paraphrase query subsets

Explanation

Correct Answer

A:

The acceptance measure focuses on full-section citation accuracy at a fixed context budget at the task or workflow boundary. Model fluency would not establish whether the integration can search fine-grained chunks then return appropriate parent spans for grounding, and it could conceal cases where returning only tiny isolated fragments removes governing context.

Incorrect Answers

B:

When source policies change daily, retired versions must stop appearing and the integration can synchronize create-update-delete events into the retrieval index; but the acceptance gate must instead inspect full-section citation accuracy at a fixed context budget to detect the targeted failure.

C:

When questions require filtering by jurisdiction, policy effective date and the integration can store reliable source metadata for filtering before similarity search; by contrast, the telecommunications support release needs evidence from full-section citation accuracy at a fixed context budget.

D:

Where users search both exact product codes and paraphrased descriptions, a relevant check would establish whether the integration can combine keyword matches with embeddings and rank the merged results; however, that proxy would not demonstrate full-section citation accuracy at a fixed context budget for the selected intervention.

Question 751

An independent assessment of a compliance assurance Claude system reports that transferring an agent identity must not silently inherit broader privileges. Which change best closes the exposed control gap?

  1. Prioritize Supervisor-mediated delegation
  2. Prioritize Independent critic with bounded revisions
  3. Prioritize Centralized permission broker for agents
  4. Prioritize Sequential dependency-aware handoffs

Explanation

Correct Answer

C:

The reported failure calls for the control under which a service-side broker verifies user authority for each proposed action. A prompt-only reminder or unrelated metric would not address the risk that transferring an agent identity must not silently inherit broader privileges. Evidence of resolution requires user-context propagation and centralized authorization checks and an improvement in privileged tool denials and policy-consistent action logs.

Incorrect Answers

A:

This addresses the failure where peer agents with full shared authority create excessive attack surface, because a supervisor assigns tasks and consolidates results within distinct permission scopes; yet it would not correct the observed weakness: transferring an agent identity must not silently inherit broader privileges.

B:

When unbounded self-critique loops can raise cost without a convergence rule, a separate evaluation pass checks defined criteria and requests capped revisions; but this remedy operates on another boundary, not the case where transferring an agent identity must not silently inherit broader privileges.

D:

When parallelizing dependent tasks can validate stale or missing artifacts, the architecture can orchestrate a staged dependency graph with explicit prerequisites; yet it would not correct the observed weakness: transferring an agent identity must not silently inherit broader privileges.

Question 752

A media licensing team is selecting the next improvement for its AI service; a long case management dialogue continually replays stale turns. What design choice has the strongest direct justification?

  1. Introduce Token-economics telemetry
  2. Introduce Memory separation by trust level
  3. Introduce Rolling conversation summary with state
  4. Introduce Exact token budget accounting
  5. Introduce Relevance-ranked context packing
  6. Introduce Source-priority truncation policy

Explanation

Correct Answer

C:

The mechanism is specific: the model configuration can preserve verified decisions and current state while pruning redundant history. A credible implementation requires explicit summary refresh and fact provenance, while its acceptance test monitors carried-forward constraint accuracy across long conversations. The alternative approaches leave the risk that raw replay increases cost and may reintroduce superseded requests.

Incorrect Answers

A:

If large context traffic causes budget spikes without obvious request-count increases, the model configuration can record uncached input, cache reads, output and tool-related usage per task; whereas the media licensing case must guard against the outcome that raw replay increases cost and may reintroduce superseded requests.

B:

When user preferences and external tool outputs enter the same conversation memory, the model configuration can store verified state separately from untrusted retrieved text and time-bound hints; this rollout still needs the following guarantee: the model configuration can preserve verified decisions and current state while pruning redundant history.

D:

When templates unexpectedly exceed request limits after tool results are appended, the model configuration can estimate input, tool-result and output allowances before sending requests; whereas the media licensing case must guard against the outcome that raw replay increases cost and may reintroduce superseded requests.

E:

When a case includes more background than the model can productively use, the model configuration can rank excerpts by task relevance and retain traceable source references; this rollout still needs the following guarantee: the model configuration can preserve verified decisions and current state while pruning redundant history.

F:

If a long research input must retain governing rules and current evidence, the model configuration can truncate lowest-priority material while protecting essential instructions and citations; the stated requirement remains unmet: the model configuration can preserve verified decisions and current state while pruning redundant history.

Question 753

A design authority for energy operations has established this requirement: failures spike only when lengthy contracts include appendices. Which intervention should be prioritized?

  1. Adopt Model capability mismatch test
  2. Adopt Context truncation investigation
  3. Adopt Retrieval recall diagnosis
  4. Adopt Overfit-to-examples analysis

Explanation

Correct Answer

B:

This control changes the relevant system boundary: the evaluation process can inspect token budgets and the presence of decisive passages after packing. The team should establish serialized prompt capture with redaction before enabling it, then evaluate missing-critical-span frequency versus context length so that increasing temperature cannot restore clipped evidence is not hidden by fluent outputs.

Incorrect Answers

A:

When complex cross-document inferences fail despite accurate retrieved passages, the evaluation process can evaluate a more capable tier on the same evidence and instructions; the alternative leaves the central requirement unproven: the evaluation process can inspect token budgets and the presence of decisive passages after packing.

C:

If generated answers omit a rule even though it exists in the knowledge base, the evaluation process can inspect whether relevant source passages reached the model; whereas the energy operations case must guard against the outcome that increasing temperature cannot restore clipped evidence.

D:

For a workload where a few-shot classifier works on familiar categories but fails novel wording, the evaluation process can test whether examples bias the model to surface patterns rather than policy rules; the alternative leaves the central requirement unproven: the evaluation process can inspect token budgets and the presence of decisive passages after packing.

Question 754

An energy operations platform team is designing a Claude-supported workflow because privacy reviewers cannot see where source documents or traces travel. Which architectural decision most directly meets that requirement?

  1. Introduce End-to-end data-flow diagram
  2. Introduce Deployment topology and residency record
  3. Introduce Prompt and model configuration manifest
  4. Introduce Decision-to-requirement traceability

Explanation

Correct Answer

A:

The proposed control fits because the delivery team can show all data stores, transformations, transfers and trust boundaries. Its production prerequisites include service inventory and data classification; check completeness of identified data-flow and access boundaries on representative tasks. Without this boundary, a high-level component box diagram can hide secondary retention paths.

Incorrect Answers

B:

If legal and platform engineers need to know which environments store data, the delivery team can document region, identity, storage and network paths for each environment; whereas the energy operations case must guard against the outcome that a high-level component box diagram can hide secondary retention paths.

C:

If multiple environments produce divergent behavior despite using one codebase, the delivery team can version prompts, model choices, tools and safety settings together; the stated scenario depends on a different mechanism: the delivery team can show all data stores, transformations, transfers and trust boundaries.

D:

When stakeholders cannot tell which architectural control satisfies which need, the delivery team can link decisions to acceptance tests and authoritative requirement IDs; that mechanism leaves another need unresolved: the delivery team can show all data stores, transformations, transfers and trust boundaries.

Question 755

A logistics coordination risk review committee is designing a Claude-supported workflow because EU customer profiling may process identifiable personal data. Which architectural decision most directly meets that requirement?

  1. Implement Privacy impact and lawful-basis review
  2. Implement HIPAA responsibility and contract assessment
  3. Implement Data residency deployment review
  4. Implement Retention and erasure workflow

Explanation

Correct Answer

A:

The intervention acts on the system behavior directly: the control layer can involve privacy counsel in lawful basis, purpose and data minimization assessment. Before rollout, verify jurisdiction mapping and data-use inventory. The relevant success test is documented lawful-basis coverage and purpose exceptions; otherwise technical encryption does not by itself establish regulatory permission to process.

Incorrect Answers

B:

If a US clinical workflow handles protected health information through vendors, the control layer can verify applicable HIPAA obligations and required business associate arrangements; that does not provide the capability actually needed: the control layer can involve privacy counsel in lawful basis, purpose and data minimization assessment.

C:

When contract terms restrict where customer records may be processed or stored, the control layer can verify regional processing, logs, backups and support access as a whole; the alternative cannot replace this key requirement: the control layer can involve privacy counsel in lawful basis, purpose and data minimization assessment.

D:

When a user requests deletion of personal data from model-related application stores, the control layer can map and process erasure across prompts, logs, caches and derived indexes as applicable; whereas the logistics coordination case must guard against the outcome that technical encryption does not by itself establish regulatory permission to process.

Question 756

A telecommunications support incident exposes a specific weakness: returning only tiny isolated fragments removes governing context. Which architectural correction addresses the cause rather than masking the symptom?

  1. Introduce Parent-child document retrieval
  2. Introduce Hybrid lexical and semantic retrieval
  3. Introduce Incremental refresh and delete propagation
  4. Introduce Semantic section-aware chunking

Explanation

Correct Answer

A:

The engineering fix changes the mechanism rather than the presentation: the integration can search fine-grained chunks then return appropriate parent spans for grounding. It requires parent-child mapping and document version IDs; a regression test should examine full-section citation accuracy at a fixed context budget in cases where returning only tiny isolated fragments removes governing context.

Incorrect Answers

B:

Where users search both exact product codes and paraphrased descriptions, the integration can combine keyword matches with embeddings and rank the merged results; yet it would not correct the observed weakness: returning only tiny isolated fragments removes governing context.

C:

When stale embeddings can support obsolete guidance after source correction, the integration can synchronize create-update-delete events into the retrieval index; but this remedy operates on another boundary, not the case where returning only tiny isolated fragments removes governing context.

D:

When fixed arbitrary cuts can detach qualifiers from the associated rule, the integration can split at meaningful sections and preserve local context around exceptions; whereas the telecommunications support defect follows the path in which returning only tiny isolated fragments removes governing context.

Question 757

The healthcare operations architect proposes omitting Centralized permission broker for agents from the production design. What is the most direct failure mode that decision would expose?

  1. The danger that unbounded self-critique loops can raise cost without a convergence rule
  2. The danger that unconstrained natural-language handoffs lose provenance and state semantics
  3. The danger that a task can exhaust resources when delegation has no bounded horizon
  4. The danger that transferring an agent identity must not silently inherit broader privileges

Explanation

Correct Answer

D:

The relevant failure mode is that transferring an agent identity must not silently inherit broader privileges. The existing design can prevent it when a service-side broker verifies user authority for each proposed action. The team needs user-context propagation and centralized authorization checks and must monitor privileged tool denials and policy-consistent action logs to support any claim of equivalent protection.

Incorrect Answers

A:

The competing hazard is that unbounded self-critique loops can raise cost without a convergence rule in a workload where drafted release notes repeatedly miss mandatory restrictions; whereas removing Centralized permission broker for agents primarily exposes the risk that transferring an agent identity must not silently inherit broader privileges.

B:

When teams merge agent outputs that refer to inconsistent record identifiers, unconstrained natural-language handoffs lose provenance and state semantics; but this is not the immediate loss associated with omitting Centralized permission broker for agents: transferring an agent identity must not silently inherit broader privileges.

C:

The competing hazard is that a task can exhaust resources when delegation has no bounded horizon in a workload where a planner sometimes delegates recursively until requests time out; however, the named safeguard addresses the distinct hazard that transferring an agent identity must not silently inherit broader privileges.

Question 758

The human-resources services go-live board accepts the premise that the model configuration can preserve verified decisions and current state while pruning redundant history. Which preparation is indispensable before it can rely on Rolling conversation summary with state operationally?

  1. Confirm correct tokenizer and reserved response budget
  2. Confirm request-level usage accounting and trace identifiers
  3. Confirm documented context priority policy
  4. Confirm explicit summary refresh and fact provenance

Explanation

Correct Answer

D:

The proposed design needs explicit summary refresh and fact provenance as its foundation. This matters because the model configuration can preserve verified decisions and current state while pruning redundant history; a plausible demonstration without the prerequisite can still leave a case where raw replay increases cost and may reintroduce superseded requests. The relevant check is carried-forward constraint accuracy across long conversations.

Incorrect Answers

A:

This prerequisite supports a method used when templates unexpectedly exceed request limits after tool results are appended, because the model configuration can estimate input, tool-result and output allowances before sending requests; but the production dependency for Rolling conversation summary with state is explicit summary refresh and fact provenance.

B:

This prerequisite supports a method used when large context traffic causes budget spikes without obvious request-count increases, because the model configuration can record uncached input, cache reads, output and tool-related usage per task; for this release, the team must first verify explicit summary refresh and fact provenance.

C:

This prerequisite supports a method used when a long research input must retain governing rules and current evidence, because the model configuration can truncate lowest-priority material while protecting essential instructions and citations; however, the named intervention needs explicit summary refresh and fact provenance before its protection can be enforced.

Question 759

A telecommunications support team is advancing Context truncation investigation beyond its proof of concept. Engineers expect that the evaluation process can inspect token budgets and the presence of decisive passages after packing. What evidence of setup readiness should be required first?

  1. Verify version-pinned model and repeatable input fixtures
  2. Verify serialized prompt capture with redaction
  3. Verify query-to-source relevance labels
  4. Verify immutable tool-result capture and lineage IDs

Explanation

Correct Answer

B:

Readiness depends on serialized prompt capture with redaction, not just a design approval. That condition supports the mechanism in which the evaluation process can inspect token budgets and the presence of decisive passages after packing. Without it, increasing temperature cannot restore clipped evidence; monitoring missing-critical-span frequency versus context length helps verify the deployed result.

Incorrect Answers

A:

The setup is useful to enable the alternative's behavior where the evaluation process can test instruction clarity while holding model and retrieval constant and the same evidence yields inconsistent formats across otherwise similar cases; for this release, the team must first verify serialized prompt capture with redaction.

C:

Where generated answers omit a rule even though it exists in the knowledge base, the evaluation process can inspect whether relevant source passages reached the model; but the production dependency for Context truncation investigation is serialized prompt capture with redaction.

D:

For the separate challenge where an agent selects correct tools but invents values absent from responses, the evaluation process can compare generated fields with raw tool result identifiers and values; for this release, the team must first verify serialized prompt capture with redaction.

Question 760

For a public-sector casework rollout, the proposed design uses End-to-end data-flow diagram, relying on the claim that the delivery team can show all data stores, transformations, transfers and trust boundaries. What enabling control must be in place to make that design enforceable?

  1. Arrange reviewed failure modes and on-call ownership
  2. Arrange immutable configuration references and release tagging
  3. Arrange infrastructure inventory and change-control ownership
  4. Arrange service inventory and data classification

Explanation

Correct Answer

D:

The required preparation is service inventory and data classification. It makes the following architecture claim enforceable: the delivery team can show all data stores, transformations, transfers and trust boundaries. If that setup is skipped, a high-level component box diagram can hide secondary retention paths, so the release should examine completeness of identified data-flow and access boundaries.

Incorrect Answers

A:

This prerequisite supports a method used when incident responders rely on the original author for every production failure, because the delivery team can document symptoms, safe mitigation, escalation and rollback procedures; however, the named intervention needs service inventory and data classification before its protection can be enforced.

B:

This prerequisite supports a method used when multiple environments produce divergent behavior despite using one codebase, because the delivery team can version prompts, model choices, tools and safety settings together; but the production dependency for End-to-end data-flow diagram is service inventory and data classification.

C:

This prerequisite supports a method used when legal and platform engineers need to know which environments store data, because the delivery team can document region, identity, storage and network paths for each environment; for this release, the team must first verify service inventory and data classification.

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