Welcome to awslagi.com – Sharing AWS, Google Cloud, Microsoft Azure IT Exams. It's free for all!

Claude Certified Architect — Professional

Advance to Claude Certified Architect — Professional (CCAR-P) with enterprise AI architecture, safety governance, and evaluation optimization.

4.8★★★★★(346 ratings)
13,980 studentsAdvanced
Validate your senior-level expertise in architecting enterprise-grade AI systems with Anthropic Claude through the Claude Certified Architect — Professional (CCAR-P) practice exam. Master complex solution design, multi-tier routing, retrieval strategies, and large-scale context engineering. This professional track emphasizes rigorous evaluation frameworks, safety and risk management, operational enablement, and executive stakeholder communication. Gain hands-on readiness for high-stakes architectural decisions and pass the CCAR-P certification with confidence.

What you'll learn

  • Enterprise Solution Design & Architecture
  • Governance, Safety & Risk Management
  • Evaluation, Testing & Cost Optimization
  • Claude Models, Prompting & Context Engineering
  • Operational Enablement & Lifecycle Management

CCAR-P

Lectures1
Questions48
Duration30 week
Students13,980

$20.00

Full Free Question Bank (48 Questions)

Claude Certified Architect — Professional (CCAR-P) Practice Questions & Verified Answers

Click any option to test your knowledge, or reveal the verified answer and detailed explanation below each question.

Question #1Part 1Solution Design & Architecture

You are evaluating a customer-support agent’s configuration. Each prospective tool belongs to one of four categories: (1) required to carry out defined tasks, (2) frequently used and reduces hand-offs, (3) occasionally helpful for unrelated work, (4) speculative future utility. Which categories should generally stay in the agent configuration?

AOnly category 3, because occasionally useful tools for unrelated work provide broader coverage and should take priority over tools required for the agent’s defined tasks.
BOnly category 4, because speculative future-utility tools provide the most flexibility and should be configured even when no defined task currently requires them.
CCategories 1 and 2 only, because they map to defined tasks and the agent’s regular hand-offs.
DAll four categories, because broader tool access is categorically better for agent performance regardless of whether the tools map to defined tasks or regular hand-offs.
Question #2Part 1Governance, Safety & Risk ManagementSelect 2

A technical team is cataloguing risks particular to Claude’s use in a document-grounded Q&A system. Which two items are failure modes inherent to LLM-based systems rather than generic software defects? (Select two.)

AAn expired TLS certificate blocks outbound API calls to Claude.
BA database connection timeout causes retrieval to return an empty result set.
CThe model refuses a legitimate query because surface features trigger an overly broad safety pattern.
DA misconfigured load balancer routes requests to a deprecated API version.
EThe model generates a plausible-sounding answer unsupported by any retrieved document.
Question #3Part 1Evaluation, Testing & Optimization

You are adding human review to a high-volume classification pipeline in which reviewing every output is impractical. Which sampling strategy best balances throughput and quality oversight?

ANo sampling, relying entirely on user complaints to reveal quality and safety problems after they affect users.
BRisk-stratified sampling that reviews all low-confidence and high-impact outputs and a smaller random sample of high-confidence routine outputs.
CInverse sampling that reviews only high-confidence routine outputs and skips low-confidence and high-impact outputs.
DUniversal review of every output regardless of confidence or throughput impact.
Question #4Part 1Evaluation, Testing & Optimization

You operate an interactive assistant where per-turn latency is the primary performance constraint. Quality on routine turns is already acceptable. Which configuration change most directly improves latency without disproportionately harming quality?

ADisable prompt caching entirely to ensure fresh context processing on every request, preventing stale prefix content from affecting latency-sensitive interactions.
BReduce retrieval depth to the top-k passages that historically cover the answer, and cache stable system-prompt content.
CIncrease retrieval depth to the corpus maximum to improve recall regardless of latency.
DSwitch every turn to the heaviest available model to maximize output quality, accepting that the increased model latency will worsen the per-turn SLO rather than improve it.
Question #5Part 1Solution Design & Architecture

You are evaluating patterns for a batch document-classification job with fixed stages: extract metadata, classify, summarize, and persist. Which pattern is the best fit, and why?

AA workflow pattern, because agentic patterns do not support tool invocations and therefore cannot execute the persist step that writes results to the downstream store.
BAn agentic pattern, because the open-ended planning capability of agents produces more consistent structured outputs than a fixed workflow graph on classification tasks.
CAn agentic pattern, because agents are categorically more accurate than workflows and therefore always preferable regardless of step predictability.
DA workflow pattern, because the steps are well-defined and predictable per-request token cost is preferred over flexibility.
Question #6Part 1Evaluation, Testing & OptimizationSelect 2

A pilot AI assistant for procurement specialists has an 89 percent first-response acceptance rate, but follow-up surveys show that specialists often override the assistant’s vendor recommendations after considering criteria the assistant did not assess. The pilot owner wants to launch the assistant unchanged because of the strong acceptance rate. Which two Discernment-competency observations should you raise **BEFORE** approving the launch? *(Select two.)*

AThe acceptance rate alone proves readiness for general production use.
BThe survey response rate may not be statistically representative of all specialists.
CThe unconsidered criteria represent a scope gap in the assistant’s input space.
DAcceptance does not establish whether recommendations remain correct after specialist review.
EThe pilot duration was probably too short to demonstrate reliability across the full year.
Question #7Part 1Solution Design & Architecture

You are a solution architect building a Claude-based assistant that has access to 60 internal tools across multiple business domains. Loading every tool definition with each request increases token usage and time to first response. Which design pattern best resolves this issue without reducing capability breadth?

AApply progressive tool discovery so a curated initial subset is exposed and additional tools are loaded on demand based on the task.
BUse a separate model call to summarize all 60 tool definitions before each user turn.
CIncrease the maximum context length and load all 60 tool definitions on every request, accepting the higher token cost and latency as necessary for full capability.
DHard-code a fixed set of five tools per request to reduce token usage, regardless of whether those tools are relevant to the current task.
Question #8Part 1Integration

You are investigating an MCP server that fails on its initial launch but works on later runs. System permission dialogs appeared during that first launch. Which response is the most appropriate?

ARecognize the first-run permission grant as the cause, document the expected behavior in onboarding guidance, and confirm that subsequent runs succeed.
BReinstall the operating system to clear all permission state without first confirming whether the one-time permission grant caused the failure.
CDisable operating-system permission dialogs entirely, accept the resulting security implications, and proceed without confirming whether the failure recurs.
DTreat the first-run failure as a permanent fault, replace the MCP server, and do not verify whether subsequent runs succeed.
Question #9Part 1Governance, Safety & Risk ManagementSelect 2

An engineering organization is rolling out Claude Code across 200 developers. A team lead proposes that AI-generated pull requests bypass the standard code-review process for changes under 50 lines, because small changes are viewed as low risk and review capacity is limited. Which two Diligence-competency objections should you raise? Each correct answer presents part of the solution. (Select two.)

ARemoving review eliminates the verification mechanism required for AI-generated output across the workflow.
BCode-review capacity should be expanded to handle every code change submitted across the team.
CThe threshold should be increased to 200 lines to capture more changes.
DAI-generated pull requests should be rejected by default across the organization.
ELine count is a poor proxy for the actual risk introduced by a code change.
Question #10Part 1Evaluation, Testing & OptimizationSelect 2

You are choosing a model for a new production workload and must complete the upstream steps before empirically testing candidate models. Which two steps must be completed **BEFORE** testing a representative sample on a candidate model? Each correct answer is part of the solution. (Select two.)

ASelect the lightest model that consistently meets the quality bar across the sample.
BRetest the model choice when a new model version becomes available.
CChoose a candidate model based on the requirements profile and known capabilities.
DDefine the quality bar, latency tolerance, and expected volume for the workload.
ESign off the production rollout plan with the platform and security teams.
Question #11Part 1Evaluation, Testing & OptimizationSelect 2

You are developing a test strategy for a Claude-based pipeline that processes sensitive financial data. Which two testing types should be prioritized to address both safety under attack and correctness across components? (Select two.)

AAdversarial tests using prompt-injection and malformed-input cases.
BRegression tests against a stable reference set of previously known-good outputs.
CSmoke tests that verify core paths after each deployment.
DIntegration tests that verify end-to-end pipeline behavior across all components.
EUnit tests targeting only individual prompt-template rendering logic.
Question #12Part 1Evaluation, Testing & OptimizationSelect 2

You are assessing instrumentation in a multi-agent system. Which two findings represent valid observability gaps in the instrumentation? Each correct answer provides a complete solution. (Select two.)

ATrace spans for each agent step are exported to the shared distributed-tracing backend.
BLatency and token usage on every span are emitted to the central metrics pipeline.
CTool-call payloads and outcomes are recorded with redaction applied to known sensitive fields.
DModel identity and version on each turn are not recorded with the turn artifacts.
ERequest-scoped correlation identifiers do not propagate across agent and tool calls.
Question #13Part 1Claude Models, Prompting & Context Engineering

A document-analysis service handles legal filings that average 80,000 tokens each. Attorneys query every filing an average of 14 times during a case. The current architecture sends the entire filing with every query. The CFO has asked you to lower per-query costs while maintaining response quality. The security officer requires that filing content not be stored outside Fabrikam's tenancy. Which optimization approach should you recommend?

ASummarize each filing once at intake and run all subsequent queries against the summary.
BCache the filing as the prompt prefix for reuse across the 14 queries per case.
CIndex filings in a vector store and retrieve only the relevant passages per query.
DMove the workload to a smaller Claude model to reduce the per-token cost paid.
Question #14Part 1Governance, Safety & Risk ManagementSelect 2

You are creating an ethics-review checklist for AI-supported deployments. Which two checks should be included on the list? Each correct answer provides a complete solution. (Select two.)

AConfirm that vendor licensing terms permit the planned production use of the model.
BVerify that outputs do not rely on generalizations about people that the underlying data does not support.
CConfirm that high-impact decisions retain human accountability rather than being attributed to the model.
DRestrict ethics review to outputs that exceed a defined model-confidence threshold.
EConfirm that latency and throughput targets are met across supported user populations.
Question #15Part 1Solution Design & Architecture

You are choosing a pattern for a compliance Q&A assistant that must respond to policy questions with citations from the authoritative internal source set. Latency, cost, and audit predictability are priorities. Which pattern is the best fit?

AAugmented LLM with retrieval-augmented generation over the indexed authoritative corpus and citation rendering on each answer.
BMulti-agent orchestration with a planner, researcher, and writer agent for every query.
CA pure agent loop with open web-browsing tools to surface the most current policy information, without constraining retrieval to the authoritative internal corpus.
DA static prompt with the entire policy corpus concatenated into every request.
Question #16Part 1Evaluation, Testing & OptimizationSelect 2

You are building an evaluation pipeline for a Claude-based deployment and must finish the specification steps before running the deployment against the dataset. Which two steps must be completed BEFORE running the deployment against the evaluation dataset? Each correct answer represents part of the solution. (Select two.)

APublish the aggregated metrics to a dashboard and gate releases on threshold checks.
BCurate and label the evaluation dataset to match the defined slices.
CReview failure cases with subject matter experts to refine the scoring rubric.
DDefine the metrics and slices the framework will report across representative, edge, and adversarial cases.
EScore the deployment outputs against the reference labels and aggregate the metrics.
Question #17Part 1Solution Design & ArchitectureSelect 2

A managed agent deployment for claims triage has expanded from 6 tools to 34 tools over 18 months as product teams added capabilities. Triage accuracy has declined from 91 percent to 78 percent, and average tool-selection latency has risen by 2.3 seconds. A junior engineer has proposed placing a tool-router agent in front of the existing agent to filter the tool list for each request. Which two findings should you present to support capability decomposition before adding the router? Each correct answer represents part of the solution. (Select two.)

ATool descriptions overlap across multiple claim categories within the agent’s tool set.
BSeveral tools have not been invoked across the most recent 90 days of traffic.
CThe 34 tools serve four distinct claim-workflow domains within the triage scope.
DThe router pattern is well documented across publicly available agent literature.
EThe proposed router introduces an additional model call on every incoming request.
Question #18Part 1Governance, Safety & Risk ManagementSelect 2

An architect is assessing a Claude-based candidate-screening tool before deployment. A stakeholder claims that no bias evaluation is needed because the model was not trained on company data. Which two responses most accurately dispute this claim? (Select two.)

ATransparency obligations are satisfied by disclosing that an AI system is in use, without further evaluation.
BFairness testing is required only when the training dataset is known to contain protected-class labels.
CThe model may carry demographic biases from pretraining that manifest in screening outcomes regardless of fine-tuning.
DThe system should be evaluated on outcome-disparity metrics across protected groups before deployment.
EBias evaluation is unnecessary when the model provider has published a responsible-use policy.
Question #19Part 1Evaluation, Testing & Optimization

You are evaluating a Claude-based system whose primary risk is undetected quality drift in safety-relevant outputs after a model-version upgrade. Which assessment activity most directly mitigates this risk?

ADisable adversarial evaluation entirely during model-version upgrade cycles to reduce evaluation cost, accepting that safety drift will go undetected until it appears in production.
BRotate adversarial inputs randomly so no two upgrades are scored on the same set.
CSkip evaluation on each model-version upgrade and rely on user-submitted complaints to surface safety drift after the upgraded model has already served production traffic.
DMaintain an adversarial evaluation set with version-attributed scoring so each upgrade is measured against the same set before promotion.
Question #20Part 1Governance, Safety & Risk ManagementSelect 2

You are conducting a risk assessment for a planned Claude-based deployment and must finish the inventory steps before evaluating threats against assets. Which two actions must be completed **before** assessing threats against assets to estimate likelihood and impact? Each correct answer represents part of the solution. *(Select two.)*

ADocument the assessment outcome with risks, mitigations, residual risk, and acceptance owners.
BRecommend mitigations and residual-risk acceptance for the risks that remain after analysis.
CIdentify the assets that the deployment touches, along with the sensitivity of each asset.
DRecommend mitigations and residual-risk acceptance for risks that remain.
EEnumerate the threat actors and attack vectors relevant to the deployment.
Question #21Part 1Evaluation, Testing & Optimization

Following an update to a prompt template, several test cases that had previously passed now generate unexpected outputs. Which test type is specifically intended to identify this kind of failure?

AIntegration tests that validate cross-component pipeline behavior.
BAdversarial tests that probe for prompt-injection vulnerabilities.
CRegression tests scored against a stable reference set of known-good behavior.
DSmoke tests that confirm high-level system availability after the change.
Question #22Part 1Governance, Safety & Risk Management

A platform team runs a self-hosted, multi-agent system on Kubernetes that coordinates seven specialized agents for invoice processing. The team uses approximately 40 percent of its engineering capacity for infrastructure maintenance, message-bus reliability, and agent-state recovery. The CFO has asked for an evaluation of a move to managed agent infrastructure to recover engineering capacity. The security officer requires all customer financial data to stay within an approved network boundary. Which factor should carry the greatest weight in your recommendation?

AWhether the current seven agents map cleanly to the patterns supported by managed agents.
BWhether managed agents reduce per-invoice token costs across the existing processing volume.
CWhether managed agents support the current bus topology used by the platform team.
DWhether managed agent data handling satisfies the network boundary required by security.
Question #23Part 1Evaluation, Testing & OptimizationSelect 2

A Claude architect at a health services organization is establishing evaluation metrics for a clinical-summary pipeline. The pipeline must stay within a per-query cost ceiling and must never expose patient data to unauthorized roles. Which two metrics directly address these requirements? (Select two.)

ABLEU score computed against a human-annotated reference summary set
BRole-based access-control enforcement rate measured on a red-team dataset
CThroughput measured as successful requests processed per minute
DPer-query token cost measured against the defined cost ceiling
EResponse latency at the 95th percentile across a one-week sample window
Question #24Part 1Evaluation, Testing & OptimizationSelect 2

A Claude-based research assistant starts generating responses that confidently contradict its retrieved source documents, even though the retrieval pipeline has not changed. Which two diagnostic actions most directly identify the root cause of this behavior? (Select two.)

AIncrease the context-window size to allow more retrieved chunks per query.
BSwitch the retrieval index to a denser embedding model to improve chunk-relevance scores.
CDetermine whether the failure reproduces on the previous model version to test for a model mismatch.
DReduce the temperature setting to lower response variance across all query types.
EInspect the system-prompt grounding instructions to determine whether citation constraints remain intact.
Question #25Part 1Claude Models, Prompting & Context Engineering

You are designing a content-moderation classifier that handles high volumes of user-generated comments under a tight per-message latency budget and uses well-defined classification labels. Which model selection best fits this workload?

AOpus, because every moderation decision requires maximum reasoning depth regardless of classification complexity.
BHaiku, because its latency and cost profile align with high-volume classification workloads that require limited reasoning depth.
CSonnet, because larger general-purpose models are preferred even when workload latency requirements are strict.
DSonnet with extended thinking enabled, because deeper reasoning should be applied to every moderation request to improve edge-case handling.
Question #26Part 1Evaluation, Testing & OptimizationSelect 2

A revenue-projection assistant is 38 percent above its monthly cost target. Profiling identifies three contributors: a 6,000-token policy preamble repeated on every call (45 percent of cost), retrieval of historical sales chunks averaging 3,000 tokens per call (30 percent), and inference on a flagship-tier model (25 percent). Stakeholders require that projection accuracy remain unchanged. Which two optimizations should you sequence first to reduce cost without affecting accuracy? Each correct answer represents part of the solution. (Select two.)

AReduce the number of historical sales chunks retrieved across each query run.
BTruncate the policy preamble to remove non-essential clauses from the prompt.
CEnable prompt caching on the static policy preamble across the recurring calls.
DSwitch the workload to a smaller, faster Claude model tier across all queries.
ECache common retrieved sales chunks accessed across many of the daily queries.
Question #27Part 1Developer Productivity & Operational EnablementSelect 2

An operations engineer says that a Claude-based pipeline started returning malformed JSON responses after a scheduled maintenance window, resulting in downstream processing failures. Which two investigative actions most directly isolate the root cause? (Select two.)

AClear the prompt cache and resubmit all pending requests to eliminate stale cached prefixes.
BCompare the current system prompt and output schema configuration against the last known good version from before the maintenance window.
CSwitch to a different Claude model tier to rule out provider-side changes as a contributing factor.
DIncrease the max_tokens limit to determine whether output truncation is causing incomplete JSON structures.
EReplay a set of premaintenance requests against the current configuration and inspect the raw model output before downstream parsing.
Question #28Part 1IntegrationSelect 2

You are assembling factors that should guide the selection among Model Context Protocol (MCP), direct API integration, and agent-to-agent handoff. Which two factors should be included on the list? Each correct answer provides a complete solution. (Select two.)

AWhether the vendor publishes detailed reference documentation for each candidate protocol.
BWhether the team has prior implementation experience with any of the candidate protocols.
CWhether the underlying transport supports encryption in transit between the integrated services.
DWhether the integration must be portable across multiple AI clients in the ecosystem.
EWhether the interaction is stateless and latency-sensitive or stateful and longer-running.
Question #29Part 1Integration

You are choosing a protocol for one low-latency, stateless tool call from a Claude-based assistant to an internal pricing service that already offers a stable HTTP API. Which integration mechanism is most appropriate?

AA direct API call to the existing endpoint with the appropriate scoped credentials.
BA long-lived stateful session protocol for a stateless single-call interaction.
CA bespoke streaming protocol layered over an unrelated asynchronous message bus.
DAn agent-to-agent handoff that introduces another Claude-based agent in front of the pricing service.
Question #30Part 1Governance, Safety & Risk ManagementSelect 2

You are assessing an integration specification for security weaknesses. Which two findings represent valid security gaps in the specification? Each correct answer provides a complete solution. (Select two.)

ATool calls execute server-side under a least-privilege service principal scoped to the requested action.
BService credentials are placed in the prompt context, where they can leak into logs and traces.
CRole-based access control is enforced only at the response-rendering layer after the model accesses restricted data.
DPer-user OAuth tokens are exchanged with scope-restricted permissions and refreshed within the active session.
ETool inputs and outputs are encrypted in transit using transport-layer security between services.
Question #31Part 1Evaluation, Testing & OptimizationSelect 2

The platform team at Trenova Systems, Inc. must lower per-query cost and p95 latency for a high-volume Claude pipeline without reducing output quality for the core use case. Which two optimizations directly address both cost and latency at the same time? (Select two.)

AEnable prompt caching on the static system-prompt prefix to reduce billable input tokens on repeated calls.
BRoute straightforward query types to a smaller, faster Claude model and reserve the full model for complex cases.
CIncrease max_tokens to reduce the frequency of truncated responses requiring follow-up calls.
DAdd a retrieval step that fetches the full source corpus into the context window before generation.
EExpand the system prompt to include additional few-shot examples on every request.
Question #32Part 1Evaluation, Testing & OptimizationSelect 2

You are implementing monitoring for a Claude-based deployment and must complete the specification steps before instrumenting the deployment. Which two steps must be completed **BEFORE** instrumenting the deployment to emit metrics and traces? Each correct answer is part of the solution. *(Select two.)*

ADefine the metrics and the slices the deployment will monitor across normal and adversarial traffic.
BTune the alert thresholds based on observed normal-state distributions to reduce false positives.
CDefine the service-level objectives (SLOs) and the error budgets the deployment will be held to.
DBuild the dashboards that surface metrics across slices at the cadence the team operates on.
EDocument the dashboards, alerts, and runbooks for the on-call rotation that will respond.
Question #33Part 1Solution Design & Architecture

During discovery, you are differentiating functional requirements from non-functional requirements. Which item is a non-functional requirement?

AThe system must extract a defined set of specific fields from invoice attachments and populate a downstream data record.
BThe system must produce a draft response that a human reviewer can edit before sending.
CThe system must classify inbound tickets into a defined set of categories.
DThe system must respond at p95 latency under 800 milliseconds at the expected request volume.
Question #34Part 1Evaluation, Testing & Optimization

A healthcare organization is assessing two Claude-powered AI architectures for a clinical-documentation assistant. Architecture X delivers higher output-quality scores, but costs $0.18 per documentation session and averages 4.2 seconds per response. Architecture Y scores slightly lower on quality metrics, but costs $0.09 per session and averages 2.1 seconds per response. The stated SLA requires responses in under 3 seconds, and the projected annual volume is 2 million documentation sessions. Which evaluation approach properly applies business-value pillar analysis to this decision?

ASelect Architecture Y based solely on the 50% cost reduction, since solution cost is the most important value pillar in healthcare budget-constrained environments.
BSelect Architecture X because the higher quality scores justify the cost premium, and any SLA gap can be addressed through infrastructure optimization after deployment.
CEliminate Architecture X on SLA grounds, then evaluate Architecture Y against the efficiency and solution cost pillars by calculating annual cost difference and assessing whether the quality delta materially impacts clinical workflow productivity.
DRecommend a hybrid approach using Architecture X for complex cases and Architecture Y for routine cases, without additional analysis, since this preserves quality where it matters most.
Question #35Part 1Claude Models, Prompting & Context Engineering

You are troubleshooting a Claude Code session in which its subagent uses 50,000 context tokens before the engineer types even a single message. Which root cause is most likely?

AThe developer's keyboard layout or input-method configuration is the cause of the elevated context consumption, introducing extra tokens before the engineer types any message.
BMany MCP servers are configured, each contributing tool definitions to the context budget; Tool Search is not enabled, so all definitions load upfront.
CThe model has internal personal preferences or default behaviors that silently consume large portions of context budget before any user message is processed, independent of tool configuration.
DThe font rendering or display-scaling settings of the IDE are converting visual output into additional context tokens, causing the high pre-session context consumption.
Question #36Part 1Integration

You are assisting an engineer whose newly added MCP server is not appearing in their Claude Code session. Which diagnostic step should be attempted first?

ADisable every MCP server in the configuration to isolate the issue, removing all tool access rather than confirming whether the new server is registered.
BSwitch the team to a different Claude product family unrelated to the existing workflow, abandoning the current configuration rather than diagnosing the registration issue.
CVerify the MCP server is registered using the listing command, then reconnect the session if needed, since configuration changes typically require a restart to take effect.
DReinstall the developer’s operating system from scratch, discarding all local configuration and tool state before attempting any targeted diagnostic steps.
Question #37Part 1Stakeholder Communication & Lifecycle ManagementSelect 2

You are identifying characteristics of strong architectural-decision communication. Which two characteristics should be included on the list? Each correct answer represents a complete solution. (Select two.)

AOwnership and review cadence for revisiting the decision when conditions or assumptions change.
BDistribution to a standing review forum with mandatory attendance from senior engineering leaders.
COmission of rejected alternatives to keep the decision document focused and concise for readers.
DAlternatives considered along with the criteria that were used to evaluate each alternative.
EUse of a consistent template across the team to standardize the visual presentation of decisions.
Question #38Part 1Governance, Safety & Risk ManagementSelect 2

A Claude architect is designing a HIPAA-compliant pipeline that handles patient records. Which two design decisions directly support HIPAA compliance requirements? (Select two.)

ASetting max_tokens to a low value to minimize the volume of text generated per request.
BSelecting the highest-capability Claude model to maximize diagnostic accuracy.
CEnforcing role-based access controls so that PHI is retrievable only by authorized personnel.
DEnsuring patient data is never included in training feedback loops sent to the model provider without a BAA in place.
EUsing streaming responses to reduce perceived latency for clinical users.
Question #39Part 1Stakeholder Communication & Lifecycle ManagementSelect 2

An architect reviews a set of architecture-documentation packages before transferring a Claude-based pipeline to an implementation team. Which two characteristics show that a documentation package is sufficient to enable implementation without continued involvement from the architect? (Select two.)

AThe document specifies integration contracts, configuration schemas, and expected input/output shapes for each component.
BThe document includes a decision log that records the rationale for key architectural choices and the alternatives rejected.
CThe document provides a high-level narrative description of the business problem without component-level detail.
DThe document includes the architect’s contact information for questions arising during implementation.
EThe document lists all Claude models that were evaluated but does not specify which was selected or why.
Question #40Part 1Developer Productivity & Operational EnablementSelect 2

The engineering lead at Trenova Systems, Inc. is assessing two proposals to improve developer workflows with Claude-assisted tooling. Proposal A integrates Claude Code into the IDE for inline code generation and review. Proposal B sends all code-generation requests through a shared Slack bot with no IDE integration. Which two observations most accurately assess these proposals against workflow-improvement objectives? (Select two.)

AProposal B is superior because centralizing requests in Slack creates an auditable log of all code-generation activity.
BProposal B introduces workflow friction by requiring developers to leave the IDE, undermining the productivity objective.
CProposal A reduces context switching by providing AI assistance at the point of development without requiring a separate tool.
DProposal B improves workflow velocity because Slack notifications create an asynchronous review queue.
EBoth proposals are equivalent because the model capabilities are identical regardless of the integration point.
Question #41Part 1Governance, Safety & Risk ManagementSelect 2

A security team is assessing two proposed controls. Control A implements an outbound tool allow-list with destination restrictions and per-call review. Control B evaluates responses against a stable adversarial evaluation set after every model-version change. Which two risk categories are correctly matched to these controls? (Select two.)

AControl A — prompt injection from adversarial content in retrieved data
BControl A — silent quality drift after a model-version upgrade
CControl A — data exfiltration via outbound tool calls
DControl B — data exfiltration via outbound tool calls
EControl B — silent quality drift after a model-version upgrade
Question #42Part 1Evaluation, Testing & Optimization

A Claude Architect reviews a post-deployment performance report for an AI-assisted legal-document summarization system. The report contains these observations: 1. Average summarization time decreased from 47 minutes to 6 minutes per document. 2. Associates spend less time on summaries, but overall billable output has not measurably changed. 3. Infrastructure costs increased by 22% because redundant retry logic generated additional API calls. 4. Some summaries require attorney correction, adding an average of 8 minutes of review per document. Which analysis correctly assigns each observation to the appropriate business-value pillar?

AObservations 1 and 2 both indicate efficiency gains; Observation 3 is a solution-cost issue; Observation 4 is a performance-SLA issue.
BObservation 1 is a transformation outcome; Observation 2 is an efficiency gain; Observation 3 is a performance-SLA degradation; Observation 4 is a solution-cost issue.
CObservations 1 and 4 together indicate a net performance-SLA improvement; Observation 2 is a transformation gap; Observation 3 is a productivity drain from over-engineering.
DObservation 1 indicates an efficiency gain; Observation 2 shows that productivity has not yet been realized; Observation 3 is a solution-cost issue; Observation 4 is an efficiency loss that partially offsets Observation 1.
Question #43Part 1Developer Productivity & Operational Enablement

A team manager needs every engineer working in the same repository to use identical MCP server definitions, without manually keeping them in sync. Which configuration approach meets this requirement?

Amanaged configuration pushed to all endpoints by the administrator
Benvironment variables set at the operating-system level on each workstation
Cproject-scope .claude/settings.json and .mcp.json files committed to the repository
Deach engineer maintains a personal ~/.claude/settings.json with the shared definitions
Question #44Part 1Governance, Safety & Risk ManagementSelect 2

A research-summarization assistant has been in production for six months. A user reports that a generated summary included a fabricated citation. The product team asks whether the incident needs architectural action or is an isolated occurrence. Which two Diligence-competency actions should you take? Each correct answer represents part of the solution. *(Select two.)*

ASample recent summaries to estimate the fabrication frequency across the population.
BDisable the assistant immediately for all current users without any prior diagnostic analysis.
CReview whether citation-grounding controls exist anywhere in the current generation pipeline.
DCommunicate to users that the assistant does not fabricate output as a matter of design.
ETreat the incident as an anecdotal isolated case and take no further investigative action.
Question #45Part 1Stakeholder Communication & Lifecycle Management

You are presenting an architecture decision to a mixed audience that includes an executive sponsor and the engineering leads responsible for implementing the decision. Which presentation approach best serves both audiences?

AOpen with a deep dive into low-level implementation details targeted at engineering leads, and stop there without addressing the business outcomes or trade-offs the executive sponsor needs.
BLead with the decision, the business outcomes it serves, and the trade-offs accepted; follow with the technical rationale, alternatives, and implementation implications for the engineering audience.
CSkip the rationale, alternatives, and trade-off discussion entirely and simply announce the chosen decision, leaving both audiences without the context needed to implement or validate it.
DPresent a single undifferentiated narrative that addresses technical and business concerns with equal weight throughout, treating both audiences as requiring the same depth on every section.
Question #46Part 1Governance, Safety & Risk ManagementSelect 2

You are identifying characteristics of robust guardrail design for an enterprise deployment. Which two characteristics should be on the list? Each correct answer provides a complete solution. (Select two.)

ACentralized log retention for guardrail violations with quarterly review by the security team.
BPer-role tool allow-lists enforced at the orchestration layer before any tool call executes.
CUser feedback channels that route reported guardrail failures into the product backlog for triage.
DPeriodic refresh of the system prompt wording to keep refusal language current and clear.
EAdversarial-input coverage in the evaluation set with regression tracking on guardrail performance.
Question #47Part 1Stakeholder Communication & Lifecycle ManagementSelect 2

You are establishing a feedback-and-alignment routine for a deployment with multiple stakeholders. Which two practices should be included in the routine? Each correct answer provides a complete solution. *(Select two.)*

ARecord each session and distribute the recording to stakeholders who could not attend live.
BEscalate any disagreement among stakeholders to the executive sponsor for binding resolution.
CRotate facilitation among the participating stakeholder groups to share ownership of the routine.
DSet a regular cadence for revisiting expectations and assumptions as conditions evolve over time.
EReconcile divergent stakeholder positions explicitly rather than papering over them in the moment.
Question #48Part 1Stakeholder Communication & Lifecycle Management

You are establishing an SLA for a Claude-based assistant. Which SLA definition is the most operationally meaningful?

AA target tied to a stakeholder sentiment measure such as “the team feels satisfied,” which cannot be measured objectively or used to trigger a documented breach response.
BA measurable target with a defined metric, threshold, evaluation window, and consequence for breach—for example, “p95 per-request latency under 800 ms over a 28-day window.”
CA qualitative commitment such as “the system will be fast and reliable,” which names no metric, threshold, or evaluation window.
DA target that names the metric and threshold but omits the evaluation window and breach consequence, leaving compliance periods and remediation triggers undefined.

📢 Unlock Full Access to Success!

"The content displayed here is a promotional preview and contains only 50% of the actual exam database. To access the complete set of updated questions, please upgrade to the Full Version."

👉 Click the button and follow the instructions to secure your full access immediately!