Anthropic CCAR-P (Claude Certified Architect - Professional) Practice Test 109

Question 1080

The human-resources services program needs a control under which the architecture can send low-risk cases to a smaller adequate model and escalate hard cases. Select TWO components of a credible implementation plan: the design decision and what must exist first. Choose TWO.

  1. Establish difficulty classifier and escalation accuracy tests
  2. Establish latency tracing and an agreed user-facing SLO
  3. Implement Availability design for downstream dependencies
  4. Implement Tiered model routing with quality gates
  5. Track fully loaded cost divided by accepted completions

Explanation

Correct Answers

A:

This step creates the enabling basis of difficulty classifier and escalation accuracy tests. That preparation supports the mechanism that the architecture can send low-risk cases to a smaller adequate model and escalate hard cases; merely choosing the control does not prove its configuration or authority is ready.

D:

This is the design control: the architecture can send low-risk cases to a smaller adequate model and escalate hard cases. It addresses the hazard that uniform highest-tier use consumes budget without equal quality benefit; the other selected element supplies the deployment foundation and, where required, the measurement of that control.

Incorrect Answers

B:

The alternative setup enables a different mechanism: the architecture can define p95 or p99 targets with workload-specific error budgets; whereas this intervention requires difficulty classifier and escalation accuracy tests as its operating precondition.

C:

This design choice is aimed at cases where Claude responses rely on a search index that sometimes fails; it is effective because the architecture can specify degraded-mode behavior and dependency-aware reliability goals. The rollout described is instead governed by the case where trivial classifications and difficult investigations share an expensive model and requires a different primary intervention; however, the intended intervention requires that the architecture can send low-risk cases to a smaller adequate model and escalate hard cases.

E:

This test would validate a different benefit: the architecture can compare total model and review cost per accepted result instead of per call; but that indicator does not verify quality-adjusted cost across routed task categories for the selected architecture.

Question 1081

Following a regression in telecommunications support, the team establishes that assuming old cache performance can distort a new cost model. Which proposed redesign most directly prevents recurrence?

  1. Enforce Cache invalidation experiments
  2. Enforce Budget-aware Skill loading
  3. Enforce Reusable governed Skill packages
  4. Enforce Small-delta prompt updates

Explanation

Correct Answer

A:

The reported failure calls for the control under which request structure changes may alter caching effectiveness. A prompt-only reminder or unrelated metric would not address the risk that assuming old cache performance can distort a new cost model. Evidence of resolution requires equivalent load tests and usage instrumentation and an improvement in cache hit and creation trends before and after configuration change.

Incorrect Answers

B:

This addresses the failure where always-on instructions spend context budget on irrelevant procedures, because the model configuration can progressively load only the approved capability needed for the task; but this remedy operates on another boundary, not the case where assuming old cache performance can distort a new cost model.

C:

Where teams repeat complex procedures with files and scripts across projects, the model configuration can package approved specialized instructions and resources for on-demand use; whereas the telecommunications support defect follows the path in which assuming old cache performance can distort a new cost model.

D:

Where a compliance team adds one new policy rule per month, the model configuration can isolate changeable instructions from stable common context; yet it would not correct the observed weakness: assuming old cache performance can distort a new cost model.

Question 1082

The defining problem for an internal audit pilot is that a repository has many documents but only a few sections matter per request. Which proposed approach addresses it most precisely?

  1. Introduce Adaptive retrieval depth
  2. Introduce Scoped tool result summarization
  3. Introduce Two-stage source navigation
  4. Introduce Context compaction after tool fan-out

Explanation

Correct Answer

C:

The mechanism is specific: the integration can locate relevant documents first and open targeted spans afterward. A credible implementation requires index and stable section pointers, while its acceptance test monitors useful-source coverage per context token spent. The alternative approaches leave the risk that supplying all document bodies creates noise and hides decisive lines.

Incorrect Answers

A:

For a workload where simple lookups need one source but investigations need deeper evidence, the integration can expand retrieval breadth only when answer sufficiency checks fail; whereas the supply-chain planning case must guard against the outcome that supplying all document bodies creates noise and hides decisive lines.

B:

When a service returns giant JSON objects when only three fields are needed, the integration can trim tool results to task-relevant validated fields before prompting; this rollout still needs the following guarantee: the integration can locate relevant documents first and open targeted spans afterward.

D:

If parallel tool results contain extensive redundant logs, the integration can aggregate key evidence and preserve original trace references instead of replaying all raw logs; whereas the supply-chain planning case must guard against the outcome that supplying all document bodies creates noise and hides decisive lines.

Question 1083

The post-incident analysis for retail operations concludes that slow manual redeployment may leave users exposed during an incident. Which remediation targets the relevant failure boundary?

  1. Apply Safe rollback and kill switch
  2. Apply Post-launch evaluation cadence
  3. Apply Vendor and version change watch
  4. Apply Operational ownership transfer

Explanation

Correct Answer

A:

The problem lies in the absence of a mechanism where the delivery team can provide reversible releases and a tested stop mechanism for side effects. Reinstating it, backed by versioned rollout mechanism and privileged emergency controls, limits the exposure that slow manual redeployment may leave users exposed during an incident; the engineering team should track containment time during simulated severe regressions.

Incorrect Answers

B:

Although the alternative reduces the risk that passing at release time cannot establish long-term reliability because the delivery team can run recurring regression tests with feedback-driven improvements; yet it would not correct the observed weakness: slow manual redeployment may leave users exposed during an incident.

C:

Where model behavior and platform capabilities can change over time, the delivery team can maintain a compatibility review for API, model and governance revisions; but this remedy operates on another boundary, not the case where slow manual redeployment may leave users exposed during an incident.

D:

When source code transfer alone does not make the service supportable, the delivery team can hand over runbooks, access, alerts and training to service owners; whereas the retail operations defect follows the path in which slow manual redeployment may leave users exposed during an incident.

Question 1084

An operations analyst for travel operations must evaluate Accessible appeal and correction path on real tasks. A test must detect whether a static model answer should not be treated as unappealable authority, even when outputs appear fluent. Which measure is appropriate?

  1. Evaluate user understanding and inappropriate reliance rate
  2. Evaluate reviewer-rated sufficiency of decision evidence
  3. Evaluate subgroup gate pass rate on each release candidate
  4. Evaluate appeal resolution time and corrected-decision tracking

Explanation

Correct Answer

D:

The release should evaluate appeal resolution time and corrected-decision tracking rather than equate successful calls with real outcomes. The control works on the premise that the control layer can offer a practical route to correction and responsible human review; the proposed measure can expose the risk that a static model answer should not be treated as unappealable authority.

Incorrect Answers

A:

The alternative indicator checks whether the control layer can label automation role, evidence basis and uncertainty at the decision point when frontline staff mistake suggestions for verified regulatory determinations; but the acceptance gate must instead inspect appeal resolution time and corrected-decision tracking to detect the targeted failure.

B:

When a customer requests reasons for a consequential recommendation, the control layer can record relevant policy criteria and source evidence with a clear decision summary; by contrast, the travel operations release needs evidence from appeal resolution time and corrected-decision tracking.

C:

When a new prompt improves overall performance but reduces accuracy for vulnerable cohorts, the control layer can block rollout unless declared subgroup safeguards remain satisfied; however, that proxy would not demonstrate appeal resolution time and corrected-decision tracking for the selected intervention.

Question 1085

During technical sign-off for education services, Drift alert with verified labels has been selected. The design premise is that the evaluation process can monitor quality proxies and periodically score sampled expert-labeled outputs. Which enabling condition must be demonstrated before production traffic is allowed?

  1. Prepare task-specific output validators and privacy-safe log structure
  2. Prepare reviewed labels and monitored cohort segmentation
  3. Prepare cost attribution and acceptance-event joining
  4. Prepare tagged tool-call spans and ownership escalation
  5. Prepare immutable configuration bundles and release audit logs

Explanation

Correct Answer

B:

The required preparation is reviewed labels and monitored cohort segmentation. It makes the following architecture claim enforceable: the evaluation process can monitor quality proxies and periodically score sampled expert-labeled outputs. If that setup is skipped, traffic-volume metrics alone cannot establish model correctness, so the release should examine labeled quality degradation detection delay.

Incorrect Answers

A:

For the separate challenge where fluent responses sometimes omit mandatory customer warnings, the evaluation process can emit categorized validation failures, not only HTTP error codes; but the production dependency for Drift alert with verified labels is reviewed labels and monitored cohort segmentation.

C:

This prerequisite supports a method used when token spend surges while resolution volume remains flat, because the evaluation process can alert on cost per accepted outcome and decompose input, output and retries; however, the named intervention needs reviewed labels and monitored cohort segmentation before its protection can be enforced.

D:

Where a billing connector intermittently fails and affects completed tasks, the evaluation process can track dependency-specific error budgets and user-facing completion; but the production dependency for Drift alert with verified labels is reviewed labels and monitored cohort segmentation.

E:

Where incidents follow a release changing prompts and tool schemas, the evaluation process can correlate failures to deployed configuration and restore a known-good bundle; for this release, the team must first verify reviewed labels and monitored cohort segmentation.

Question 1086

A design authority for human-resources services has established this requirement: traffic spikes exceed sustainable interactive capacity. Which intervention should be prioritized?

  1. Deploy Cost per accepted outcome budgeting
  2. Deploy Tiered model routing with quality gates
  3. Deploy Capacity forecast from token demand
  4. Deploy Load-shedding with graceful fallback

Explanation

Correct Answer

D:

The architectural distinction is that the architecture can reject or queue nonessential work while preserving critical requests. The decision should be implemented with priority tiers and admitted-concurrency limits and evaluated against critical-request success rate during overload drills; otherwise, unbounded concurrency can collapse latency for every tenant.

Incorrect Answers

A:

For a workload where a cheap model yields many outputs that reviewers reject, the architecture can compare total model and review cost per accepted result instead of per call; the missing operational behavior is specific: the architecture can reject or queue nonessential work while preserving critical requests.

B:

When trivial classifications and difficult investigations share an expensive model, the architecture can send low-risk cases to a smaller adequate model and escalate hard cases; the current decision instead depends on a specific capability: the architecture can reject or queue nonessential work while preserving critical requests.

C:

For a workload where requests differ sharply in input length and tool loops, the architecture can forecast using token volume, concurrency and retry behavior rather than call count alone; whereas the human-resources services case must guard against the outcome that unbounded concurrency can collapse latency for every tenant.

Question 1087

The manufacturing quality go-live board accepts the premise that the integration can locate relevant documents first and open targeted spans afterward. Which preparation is indispensable before it can rely on Two-stage source navigation operationally?

  1. Prepare document index and stable section pointers
  2. Prepare calibrated sufficiency criteria and retrieval budget
  3. Prepare field-level projection rules and schema checking
  4. Prepare searchable catalog with trusted metadata

Explanation

Correct Answer

A:

Before deployment the team must confirm index and stable section pointers. Without that preparation, the claim that the integration can locate relevant documents first and open targeted spans afterward may remain unenforceable; the related exposure is that supplying all document bodies creates noise and hides decisive lines. The acceptance indicator is useful-source coverage per context token spent.

Incorrect Answers

B:

This prerequisite supports a method used when simple lookups need one source but investigations need deeper evidence, because the integration can expand retrieval breadth only when answer sufficiency checks fail; but the production dependency for Two-stage source navigation is index and stable section pointers.

C:

The setup is useful to enable the alternative's behavior where the integration can trim tool results to task-relevant validated fields before prompting and a service returns giant JSON objects when only three fields are needed; for this release, the team must first verify index and stable section pointers.

D:

Where dozens of APIs have large OpenAPI specifications, the integration can retrieve only candidate tool schemas after coarse capability discovery; however, the named intervention needs index and stable section pointers before its protection can be enforced.

Question 1088

The logistics coordination project must justify retaining Cache invalidation experiments to a skeptical review committee. What specific failure does this control prevent?

  1. A failure where stale unversioned snippets can survive a policy revision
  2. A failure where always-on instructions spend context budget on irrelevant procedures
  3. A failure where changing the cached prefix can invalidate reuse and trigger new writes
  4. A failure where copying long procedures into every chat creates drift and context bloat
  5. A failure where assuming old cache performance can distort a new cost model

Explanation

Correct Answer

E:

The risk is not a generic model inconvenience: assuming old cache performance can distort a new cost model. The proposed safeguard changes that boundary because request structure changes may alter caching effectiveness, and the release would need equivalent load tests and usage instrumentation plus cache hit and creation trends before and after configuration change to show the risk remains contained.

Incorrect Answers

A:

If multiple teams repeatedly consult the same policy edition, stale unversioned snippets can survive a policy revision; but this is not the immediate loss associated with omitting Cache invalidation experiments: assuming old cache performance can distort a new cost model.

B:

If dozens of seldom-used capabilities are advertised on every turn, always-on instructions spend context budget on irrelevant procedures; however, the named safeguard addresses the distinct hazard that assuming old cache performance can distort a new cost model.

C:

When many requests reuse lengthy identical policy instructions, changing the cached prefix can invalidate reuse and trigger new writes; whereas removing Cache invalidation experiments primarily exposes the risk that assuming old cache performance can distort a new cost model.

D:

When teams repeat complex procedures with files and scripts across projects, copying long procedures into every chat creates drift and context bloat; but this is not the immediate loss associated with omitting Cache invalidation experiments: assuming old cache performance can distort a new cost model.

Question 1089

A field-service coordination service is moving to production with this aim: the delivery team can provide reversible releases and a tested stop mechanism for side effects. Choose TWO complementary choices, an architectural control and the precondition it depends on. Choose TWO.

  1. Establish staffed operating team and escalation agreement
  2. Implement Safe rollback and kill switch
  3. Implement Continuous cost-governance review
  4. Establish versioned rollout mechanism and privileged emergency controls
  5. Track stage exit criteria satisfied before broader rollout

Explanation

Correct Answers

B:

The control itself makes the following capability possible: the delivery team can provide reversible releases and a tested stop mechanism for side effects. That matters when slow manual redeployment may leave users exposed during an incident; it does not replace the required enabling condition or outcome test.

D:

This step creates the enabling basis of versioned rollout mechanism and privileged emergency controls. That preparation supports the mechanism that the delivery team can provide reversible releases and a tested stop mechanism for side effects; merely choosing the control does not prove its configuration or authority is ready.

Incorrect Answers

A:

The alternative setup enables a different mechanism: the delivery team can hand over runbooks, access, alerts and training to service owners; whereas this intervention requires versioned rollout mechanism and privileged emergency controls as its operating precondition.

C:

This design choice is aimed at cases where adoption growth changes the mix of tokens, tools and support effort; it is effective because the delivery team can reassess spend versus accepted outcomes on a recurring schedule; however, the intended intervention requires that the delivery team can provide reversible releases and a tested stop mechanism for side effects.

E:

This test would validate a different benefit: the delivery team can require explicit discovery, design, pilot, hardening and operations gates; but that indicator does not verify containment time during simulated severe regressions for the selected architecture.

Question 1090

The post-incident analysis for media licensing concludes that a static model answer should not be treated as unappealable authority. Which remediation targets the relevant failure boundary?

  1. Deploy Subgroup error disparity evaluation
  2. Deploy Fairness regression release gate
  3. Deploy Accessible appeal and correction path
  4. Deploy Bias-aware dataset curation
  5. Deploy Transparency on assistant limitations
  6. Deploy Explainable evidence trail

Explanation

Correct Answer

C:

This intervention addresses the failure at its source because the control layer can offer a practical route to correction and responsible human review. To keep the issue from recurring, establish assigned appeal owners and traceable case identifiers and verify appeal resolution time and corrected-decision tracking; otherwise a static model answer should not be treated as unappealable authority can persist.

Incorrect Answers

A:

When pooled averages can conceal unequal outcomes, the control layer can test quality and harm rates separately for relevant populations; yet it would not correct the observed weakness: a static model answer should not be treated as unappealable authority.

B:

This addresses the failure where optimizing only mean accuracy can impose concentrated harm, because the control layer can block rollout unless declared subgroup safeguards remain satisfied; but this remedy operates on another boundary, not the case where a static model answer should not be treated as unappealable authority.

D:

Where historical training examples encode legacy operational disparities, the control layer can audit label quality and group representation before using examples for evaluation; yet it would not correct the observed weakness: a static model answer should not be treated as unappealable authority.

E:

Although the alternative reduces the risk that a hidden disclaimer elsewhere does not prevent false assumptions in workflow because the control layer can label automation role, evidence basis and uncertainty at the decision point; but this remedy operates on another boundary, not the case where a static model answer should not be treated as unappealable authority.

F:

Where a customer requests reasons for a consequential recommendation, the control layer can record relevant policy criteria and source evidence with a clear decision summary; whereas the media licensing defect follows the path in which a static model answer should not be treated as unappealable authority.

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