Anthropic CCAR-P (Claude Certified Architect - Professional) Practice Test 100

Question 990

A software platform architecture gate expects the following behavior: the control layer can keep traceable control evidence, access decisions and versioned approvals. Which TWO actions together supply the design intervention and its operating dependency? Choose TWO.

  1. Implement Retention and erasure workflow
  2. Track minimization compliance and utility impact by field
  3. Establish contractual residency definition and vendor processing disclosures
  4. Implement Regulatory evidence package
  5. Establish controlled retention and immutable evidence identifiers

Explanation

Correct Answers

D:

The architecture component is justified by its behavior: the control layer can keep traceable control evidence, access decisions and versioned approvals. It targets the exposure that an informal assurance statement cannot reconstruct a production event, while the complementary selected actions provide readiness and, for a three-part package, evidence.

E:

Choosing this setup item establishes controlled retention and immutable evidence identifiers, which is needed for reliable operation when the control layer can keep traceable control evidence, access decisions and versioned approvals. The Regulatory evidence package rollout should not be declared ready until controlled retention and immutable evidence identifiers is verified in its operating environment.

Incorrect Answers

A:

This design choice is aimed at cases where a user requests deletion of personal data from model-related application stores; it is effective because the control layer can map and process erasure across prompts, logs, caches and derived indexes as applicable. The rollout described is instead governed by the case where auditors need to see when governance controls were enforced for decisions and requires a different primary intervention; however, the intended intervention requires that the control layer can keep traceable control evidence, access decisions and versioned approvals.

B:

This test would validate a different benefit: the control layer can transform and suppress unnecessary personal fields before invoking tools; but that indicator does not verify audit evidence completeness across sampled cases for the selected architecture.

C:

The alternative setup enables a different mechanism: the control layer can verify regional processing, logs, backups and support access as a whole; whereas this intervention requires controlled retention and immutable evidence identifiers as its operating precondition.

Question 991

For a healthcare operations rollout, the proposed design uses Inference request batching where latency permits, relying on the claim that the evaluation process can group asynchronous requests using the available batch workflow. What enabling control must be in place to make that design enforceable?

  1. Prepare independence proof and capacity testing
  2. Prepare relevance evaluation and source-preserving deduplication
  3. Prepare durable reconciliation and batch failure handling
  4. Prepare dependency importance rules and explicit fallback messages
  5. Prepare classifier calibration and tier-specific test coverage

Explanation

Correct Answer

C:

The required preparation is durable reconciliation and batch failure handling. It makes the following architecture claim enforceable: the evaluation process can group asynchronous requests using the available batch workflow. If that setup is skipped, interactive per-item calls waste overhead for delay-tolerant workloads, so the release should examine finished items per dollar within the batch window.

Incorrect Answers

A:

For the separate challenge where three independent classifications currently run sequentially, the evaluation process can parallelize work while joining outputs under a bounded concurrency cap; however, the named intervention needs durable reconciliation and batch failure handling before its protection can be enforced.

B:

The setup is useful to enable the alternative's behavior where the evaluation process can deduplicate and rank passages before model invocation and redundant retrieved passages consume most of the prompt budget; but the production dependency for Inference request batching where latency permits is durable reconciliation and batch failure handling.

D:

The setup is useful to enable the alternative's behavior where the evaluation process can use per-dependency deadlines and continue only if required evidence is available and a task waits for a slow optional analytics integration; however, the named intervention needs durable reconciliation and batch failure handling before its protection can be enforced.

E:

The setup is useful to enable the alternative's behavior where the evaluation process can raise model capacity selectively after evidence-based difficulty checks and lightweight answers work on simple cases but fail specialized regulatory reasoning; but the production dependency for Inference request batching where latency permits is durable reconciliation and batch failure handling.

Question 992

The public-sector casework delivery group must make the following claim executable: the architecture can separate conflict detection and resolution from answer drafting. Which TWO measures pair the right intervention with the right readiness condition? Choose TWO.

  1. Establish source precedence rules and escalation ownership
  2. Implement Exception-first task splitting
  3. Establish a source inventory and per-source retrieval contracts
  4. Track failed handoffs caused by missing upstream artifacts
  5. Implement Reconciliation of conflicting sources

Explanation

Correct Answers

A:

This step creates the enabling basis of source precedence rules and escalation ownership. That preparation supports the mechanism that the architecture can separate conflict detection and resolution from answer drafting; merely choosing the control does not prove its configuration or authority is ready.

E:

Implementing the intervention changes the operational mechanism so that the architecture can separate conflict detection and resolution from answer drafting. The failure to prevent is that direct summarization may arbitrarily select one side without noting dispute; configuration and validation remain separate release obligations.

Incorrect Answers

B:

This design choice is aimed at cases where typical requests succeed but rare regulatory exceptions cause serious errors; it is effective because the architecture can separate normal and exception branches with stronger checks for the latter; however, the intended intervention requires that the architecture can separate conflict detection and resolution from answer drafting.

C:

The alternative setup enables a different mechanism: the architecture can gather and validate evidence before asking Claude to draw conclusions; whereas this intervention requires source precedence rules and escalation ownership as its operating precondition.

D:

This test would validate a different benefit: the architecture can model the task as a DAG and execute nodes only when prerequisites are satisfied; but that indicator does not verify conflicts detected and resolved with provenance for the selected architecture.

Question 993

An operations analyst for energy operations must evaluate Agent-to-agent task delegation on real tasks. A test must detect whether sharing unrestricted tool execution blurs service ownership and trust boundaries, even when outputs appear fluent. Which measure is appropriate?

  1. Measure contract compliance and error handling at the service edge
  2. Measure event-to-action latency and duplicate-event rate
  3. Measure delegation completion and verified artifact handoffs
  4. Measure authorized tool compatibility and change success across clients

Explanation

Correct Answer

C:

Tracking delegation completion and verified artifact handoffs provides direct evidence that the system delivers its intended effect: the integration can communicate through explicit task and artifact contracts between agents. The measure is more informative than prompt counts because the failure being measured is that sharing unrestricted tool execution blurs service ownership and trust boundaries.

Incorrect Answers

A:

The alternative indicator checks whether the integration can call the model through the application service with controlled retries and validation when a backend requires tightly scoped Claude inference inside its existing request pipeline; however, that proxy would not demonstrate delegation completion and verified artifact handoffs for the selected intervention.

B:

The alternative indicator checks whether the integration can accept validated inbound events and enqueue bounded downstream tasks when new repository events should start work without constant polling; but the acceptance gate must instead inspect delegation completion and verified artifact handoffs to detect the targeted failure.

D:

Where many Claude clients need access to the same governed business capability, a relevant check would establish whether the integration can expose reusable typed tools and resources through a controlled MCP server; however, that proxy would not demonstrate delegation completion and verified artifact handoffs for the selected intervention.

Question 994

A connector occasionally returns an invalid response object, yet the team proposes only changing system-prompt wording to handle the failures. Which limitation makes tool-call failure isolation necessary?

  1. The risk that top-level request durations cannot identify the responsible dependency
  2. The risk that guessing from one screenshot hides true upstream causes
  3. The risk that a one-off patch without tests allows the same bug to return
  4. The risk that changing the system prompt cannot repair invalid connector payloads

Explanation

Correct Answer

D:

The risk is not a generic model inconvenience: changing the system prompt cannot repair invalid connector payloads. The proposed safeguard changes that boundary because the engineering workflow can inspect typed tool errors and validate provider-response contracts, and the release would need captured call IDs and integration test harness plus error resolution and schema compliance by tool version to show the risk remains contained.

Incorrect Answers

A:

If a slow multi-tool run obscures its longest step, top-level request durations cannot identify the responsible dependency; whereas removing Tool-call failure isolation primarily exposes the risk that changing the system prompt cannot repair invalid connector payloads.

B:

The competing hazard is that guessing from one screenshot hides true upstream causes in a workload where intermittent agent bugs receive speculative prompt fixes without fixtures; but this is not the immediate loss associated with omitting Tool-call failure isolation: changing the system prompt cannot repair invalid connector payloads.

C:

If a production agent repeatedly reintroduces a previously fixed defect, a one-off patch without tests allows the same bug to return; however, the named safeguard addresses the distinct hazard that changing the system prompt cannot repair invalid connector payloads.

Question 995

The field-service coordination deployment uses Modular prompt composition, but executive reports currently count only successful model responses. The residual concern is that shipping all instructions to every request increases interference and token cost. Which acceptance measure would reveal real progress?

  1. Evaluate regression rate and context churn per rule update
  2. Evaluate task compliance and token usage per assembled module set
  3. Evaluate task completion and version adoption for the approved Skill
  4. Evaluate cache hit and creation trends before and after configuration change

Explanation

Correct Answer

B:

A defensible evaluation looks at task compliance and token usage per assembled module set. The control's causal claim is that the model configuration can assemble versioned role-specific modules for the active task. Monitoring that outcome, not merely elapsed tokens or success-looking text, helps catch shipping all instructions to every request increases interference and token cost.

Incorrect Answers

A:

Where a compliance team adds one new policy rule per month, a relevant check would establish whether the model configuration can isolate changeable instructions from stable common context; by contrast, the field-service coordination release needs evidence from task compliance and token usage per assembled module set.

C:

The proposed measure would suit a setting where teams repeat complex procedures with files and scripts across projects, since the model configuration can package approved specialized instructions and resources for on-demand use; but the acceptance gate must instead inspect task compliance and token usage per assembled module set to detect the targeted failure.

D:

Where a rollout changes tool definitions alongside cached prompts, a relevant check would establish whether the model configuration can test hit behavior because request structure changes may alter caching effectiveness; by contrast, the field-service coordination release needs evidence from task compliance and token usage per assembled module set.

Question 996

An energy operations product engineering team is designing a Claude-supported workflow because a prototype team leaves without trained on-call operators. Which architectural decision most directly meets that requirement?

  1. Implement Stage-gated delivery lifecycle
  2. Implement Vendor and version change watch
  3. Implement Post-launch evaluation cadence
  4. Implement Operational ownership transfer

Explanation

Correct Answer

D:

The proposed control fits because the delivery team can hand over runbooks, access, alerts and training to service owners. Its production prerequisites include approved enabling configuration and accountable owner; check handoff drill completion and accepted ownership checklists on representative tasks. Without this boundary, source code transfer alone does not make the service supportable.

Incorrect Answers

A:

When a team attempts global launch before measuring pilot outcomes, the delivery team can require explicit discovery, design, pilot, hardening and operations gates; the missing operational behavior is specific: the delivery team can hand over runbooks, access, alerts and training to service owners.

B:

When model behavior and platform capabilities can change over time, the delivery team can maintain a compatibility review for API, model and governance revisions; the current decision instead depends on a specific capability: the delivery team can hand over runbooks, access, alerts and training to service owners.

C:

When accuracy drifts gradually as policy sources and user behavior change, the delivery team can run recurring regression tests with feedback-driven improvements; whereas the energy operations case must guard against the outcome that source code transfer alone does not make the service supportable.

Question 997

The defining problem for an energy operations pilot is that a service appears accurate overall but error rates differ across language groups. Which proposed approach addresses it most precisely?

  1. Deploy Subgroup error disparity evaluation
  2. Deploy Accessible appeal and correction path
  3. Deploy Bias-aware dataset curation
  4. Deploy Transparency on assistant limitations

Explanation

Correct Answer

A:

The intervention acts on the system behavior directly: the control layer can test quality and harm rates separately for relevant populations. Before rollout, verify lawful representative evaluation data and subgroup definitions. The relevant success test is false-negative and false-positive gaps by subgroup; otherwise pooled averages can conceal unequal outcomes.

Incorrect Answers

B:

For a workload where people affected by suggestions need to challenge inaccurate case records, the control layer can offer a practical route to correction and responsible human review; whereas the energy operations case must guard against the outcome that pooled averages can conceal unequal outcomes.

C:

When historical training examples encode legacy operational disparities, the control layer can audit label quality and group representation before using examples for evaluation; this rollout still needs the following guarantee: the control layer can test quality and harm rates separately for relevant populations.

D:

When frontline staff mistake suggestions for verified regulatory determinations, the control layer can label automation role, evidence basis and uncertainty at the decision point; the stated requirement remains unmet: the control layer can test quality and harm rates separately for relevant populations.

Question 998

The enterprise knowledge management release team needs a numerical gate for Inference request batching where latency permits. Rather than equating deployment with success, it wants to catch cases in which interactive per-item calls waste overhead for delay-tolerant workloads. Which indicator should it track?

  1. Monitor accepted outcome cost by escalation bucket
  2. Monitor quality-adjusted total cost under each candidate
  3. Monitor retained answer support per input token
  4. Monitor finished items per dollar within the batch window

Explanation

Correct Answer

D:

The acceptance measure focuses on finished items per dollar within the batch window at the task or workflow boundary. Model fluency would not establish whether the evaluation process can group asynchronous requests using the available batch workflow, and it could conceal cases where interactive per-item calls waste overhead for delay-tolerant workloads.

Incorrect Answers

A:

Where lightweight answers work on simple cases but fail specialized regulatory reasoning, a relevant check would establish whether the evaluation process can raise model capacity selectively after evidence-based difficulty checks; by contrast, the enterprise knowledge management release needs evidence from finished items per dollar within the batch window.

B:

When shaving tokens degrades downstream extraction precision, the evaluation process can compare token savings against full correction and review cost; however, that proxy would not demonstrate finished items per dollar within the batch window for the selected intervention.

C:

The proposed measure would suit a setting where redundant retrieved passages consume most of the prompt budget, since the evaluation process can deduplicate and rank passages before model invocation; but the acceptance gate must instead inspect finished items per dollar within the batch window to detect the targeted failure.

Question 999

The defining problem for an energy operations pilot is that a cheap model yields many outputs that reviewers reject. Which proposed approach addresses it most precisely?

  1. Deploy Commercial stage-gate on measurable ROI
  2. Deploy Tiered model routing with quality gates
  3. Deploy Percentile latency SLO design
  4. Deploy Cost per accepted outcome budgeting

Explanation

Correct Answer

D:

Under the stated constraints, the architecture can compare total model and review cost per accepted result instead of per call. That directly addresses the operational failure without relying on a generic prompt reminder. Production depends on tagged cost accounting and a quality acceptance rubric; the result to monitor is fully loaded cost divided by accepted completions.

Incorrect Answers

A:

For a workload where a sponsor requests broad deployment without benefit evidence, the architecture can bind rollout gates to accepted outcomes, user adoption and total operating cost; the stated scenario depends on a different mechanism: the architecture can compare total model and review cost per accepted result instead of per call.

B:

For a workload where 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; that mechanism leaves another need unresolved: the architecture can compare total model and review cost per accepted result instead of per call.

C:

When average response time looks fine while urgent users face long delays, the architecture can define p95 or p99 targets with workload-specific error budgets; whereas the energy operations case must guard against the outcome that token-price comparisons alone ignore rework and failure rates.

Question 1000

A contract administration release candidate fails review after assessors observe that sharing unrestricted tool execution blurs service ownership and trust boundaries. What intervention should take priority?

  1. Deploy Server-side business API gateway
  2. Deploy CLI invocation in controlled automation
  3. Deploy Agent-to-agent task delegation
  4. Deploy MCP server as standardized tool boundary

Explanation

Correct Answer

C:

This is the causal correction: the integration can communicate through explicit task and artifact contracts between agents. With identity propagation and contract versioning in place, the team can test delegation completion and verified artifact handoffs and show that the problematic case in which sharing unrestricted tool execution blurs service ownership and trust boundaries has been addressed.

Incorrect Answers

A:

When embedding policy separately in each prompt creates inconsistent controls, the integration can place governed APIs behind one enforcement and observability boundary; yet it would not correct the observed weakness: sharing unrestricted tool execution blurs service ownership and trust boundaries.

B:

Where a development workflow needs a reproducible one-off command in CI, the integration can use a noninteractive CLI process with bounded permissions and captured exit status; but this remedy operates on another boundary, not the case where sharing unrestricted tool execution blurs service ownership and trust boundaries.

D:

When duplicating bespoke connectors across clients multiplies maintenance, the integration can expose reusable typed tools and resources through a controlled MCP server; yet it would not correct the observed weakness: sharing unrestricted tool execution blurs service ownership and trust boundaries.

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