Anthropic CCAR-P Claude Certified Architect - Professional Exam Practice Test

Documentation remains grounded only when the generating agent can inspect the authoritative implementation. A purpose-built subagent should receive scoped filesystem and code-search capabilities, identify the relevant modules, interfaces, configuration files, tests, and call sites, and generate documentation from that evidence. This approach minimizes reliance on training-data recall, which cannot represent the repository’s current private code or recent changes.

Subagents are useful because repository exploration occurs in their own context, allowing them to return focused findings or proposed edits without filling the main conversation with every file read. Their tool access should follow least privilege: read access to required paths, narrowly controlled write access, and explicit exclusions for secrets or unrelated repositories.

Generated documentation should enter the team’s established change-control process as a patch, branch, or pull request. Normal review, automated checks, link validation, example execution, and human approval provide accountability and prevent unverified material from being published. Direct publication bypasses these safeguards, while disabling repository tools makes factual grounding impossible.

Study Guide references/topics: [Claude Code custom subagents](https://docs.anthropic.com/en/docs/claude-code/sub-agents); [codebase exploration workflows] (https://docs.anthropic.com/en/docs/claude-code/common-workflows); source-grounded generation; least-privilege tools; human review and lifecycle controls.

A constrained retrieval-augmented generation architecture best satisfies the grounding, citation, latency, cost, and audit requirements. The system first indexes the approved internal policy corpus. At request time, it retrieves a small set of relevant passages, supplies those passages to Claude with stable source identifiers, and requires the answer to cite the supporting material. Anthropic’s search-result content blocks are specifically designed for this pattern: they allow custom RAG systems to provide source-attributed internal content and enable Claude to render citations against those results.

The retrieval boundary is critical. Only approved and version-controlled policy sources should be searchable, with document identifiers, effective dates, access controls, retrieval scores, and cited passages retained for audit reconstruction. This produces substantially more predictable evidence than unconstrained web research.

A planner-researcher-writer system introduces unnecessary model calls, orchestration complexity, cost, latency, and behavioral variance for routine policy lookup. Open-web browsing violates the authoritative-source requirement and increases prompt-injection and provenance risk. Concatenating the complete corpus into every prompt wastes tokens, increases latency, and can reduce retrieval precision as the corpus expands.

Study Guide references/topics: Search-result blocks for cited RAG; Claude citations; authoritative-corpus retrieval; source attribution; auditable context engineering.

===============

Authorization and compliance logging must remain enforceable application-layer controls. Option B requires the application to authenticate the requester, resolve the user’s role and entitlements, and deny the operation when the user lacks authority. A Skill can instruct Claude how to behave, but it is not a reliable security boundary and cannot replace identity-based access control enforced by the surrounding system.

Option A similarly belongs at the application layer. The application must create an authoritative audit event containing the requester, timestamp, Skill and model version, relevant input or content identifiers, generated output, review status, and retention classification. Audit evidence cannot depend on whether the model remembers or chooses to invoke a logging instruction.

Options C, D, and E describe reusable generation behavior that appropriately belongs within the approved Skill: house style, the standardized disclaimer, and the prohibited-language policy. The application may still validate their presence after generation, but the Skill remains the primary reusable content package.

The correct architecture separates behavioral guidance from deterministic enforcement. Skills supply domain instructions and templates; the application enforces identity, authorization, audit retention, data access, output validation, and consequential-action controls.

Study Guide references/topics: Skills versus application controls; authentication; authorization; compliance logging; deterministic guardrails; policy enforcement boundaries.

===============

An evaluation set must represent the populations, operating conditions, and use cases that the deployed system will encounter. A dataset drawn from only one geographic region and one tenure band cannot reliably demonstrate that the hiring-support tool performs consistently across its intended users. Expanding the evaluation set enables architects to identify regional performance differences, tenure-related errors, unsupported generalizations, and potentially discriminatory outcomes before broader deployment. The system should then be rescored using the expanded dataset, with results examined across relevant population slices rather than relying only on an aggregate score. Options A and B preserve or worsen the representation gap. Option D is also inappropriate because qualitative feedback can supplement structured evaluation but cannot replace measurable, repeatable testing of accuracy, fairness, and reliability across the intended deployment population.

Statement 1 measures efficiency because the same operating unit—one procurement analyst—now processes three times as many purchase orders per day. It reflects increased throughput and reduced effort per transaction within the existing procurement process.

Statement 2 represents transformation. The organization has not merely accelerated manual data entry; it has eliminated that role from the workflow and redirected employees toward higher-value vendor relationship management. This is a structural redesign of responsibilities and operating practices.

Statement 3 is a solution-cost issue because the measured API integration expense exceeds the approved per-transaction financial model by 40%. The architect must investigate token consumption, model routing, caching, retrieval overhead, tool calls, and integration charges while protecting the required quality level.

Statement 4 directly concerns the performance SLA. The observed 11-second latency is compared with an explicit contractual or operational target of five seconds. This is not simply an efficiency complaint; it is a measurable breach of a defined service-level commitment.

The architect should map each stakeholder statement to its dominant value pillar before recommending remediation because throughput, organizational transformation, cost control, and SLA compliance require different architectural responses.

Study Guide references/topics: Business-value pillars; efficiency; transformation; solution cost; performance SLAs; stakeholder-outcome classification.

===============

Federated authorship allows each domain team to maintain the procedural knowledge it understands while preserving organization-wide trust through a central publication gate. Team authors can update coding standards, review practices, and postmortem procedures without waiting for a single platform group to write every change. Central reviewers can validate ownership, security, tool permissions, duplication, versioning, testing evidence, and compliance before publication. Per-developer or fully decentralized publication creates inconsistent and potentially unsafe Skills. Fully centralized authorship becomes a scalability bottleneck and separates content maintenance from domain expertise. Anthropic describes Skills as packages of instructions, scripts, and resources and warns that repository Skills may grant broad tool access, making formal review essential. Claude Code Skills

Responsible deployment requires users to receive an accurate representation of the system with which they are interacting. Option C provides that transparency while also establishing a practical escalation path. Disclosure should be proportionate to the material role AI plays, expressed in language appropriate to the audience, and aligned with organizational policy and applicable regulatory obligations.

A documented human-contact route is important when the assistant cannot resolve an issue, produces a disputed result, encounters a high-impact exception, or handles a matter requiring human authority. Anthropic’s Responsible Scaling Policy recognizes escalation to human reviewers as an appropriate safeguard for edge cases and situations requiring human judgment. Responsible Scaling Policy

Options A and D intentionally conceal material AI involvement from affected users, weakening informed decision-making, auditability, and trust. Option B is also unsuitable because transparency does not require revealing proprietary prompts, confidential controls, or internal security mechanisms; however, it does require an honest explanation of AI involvement and the system’s operational role.

The correct design combines disclosure, understandable limitations, human escalation, and records demonstrating that the organization’s transparency requirements are consistently applied.

Study Guide references/topics: AI transparency; informed user interaction; human escalation; responsible deployment; disclosure policy; operational accountability.

===============

An Architecture Decision Record is designed to preserve a significant architectural choice in a concise, durable, and reviewable form. The context explains the problem, constraints, and forces that shaped the decision. The decision section records the selected approach. Alternatives demonstrate that other viable options were evaluated, while consequences identify benefits, costs, limitations, operational obligations, and residual risks. The date and authors establish accountability and historical sequence.

This structure supports audit because a reviewer can trace why a model, integration pattern, retrieval architecture, security boundary, or deployment approach was chosen. It also supports onboarding by preventing new team members from interpreting intentional design choices as arbitrary implementation details.

A slide deck may communicate a decision at a meeting but is weak as the sole authoritative record when it lacks written rationale. A code comment is too localized and may not capture system-level implications or rejected alternatives. A verbal conversation creates no durable evidence and depends entirely on individual memory.

The ADR should be version-controlled, linked to affected requirements and implementation artifacts, assigned an owner, and reviewed when its assumptions or operating conditions materially change.

Study Guide references/topics: Architecture Decision Records; decision context; alternatives; consequences; auditability; onboarding; lifecycle review.

===============

The observed failure is a groundedness defect: the response contains claims that cannot be traced to retrieved evidence. Option A directly constrains generation to the supplied sources, requires visible attribution, and verifies that each material claim has supporting text. Unsupported claims should be removed, qualified, or replaced with an explicit statement that the available sources are insufficient. Regression evaluation in Option B may detect the failure but does not itself prevent unsupported claims. OAuth addresses identity and authorization, while input classifiers address malicious or unsafe retrieved content; neither verifies factual support. Anthropic recommends citations, supporting quotations, and post-generation claim verification to reduce hallucinations in document-grounded applications. Reducing hallucinations

A direct call to the existing stable HTTP endpoint is the simplest mechanism that satisfies the stated requirements. The interaction is stateless, requires only one tool invocation, and has a strict latency objective. Adding session management, message-bus translation, or another model-mediated agent would introduce unnecessary network hops, operational dependencies, failure modes, and processing latency.

The assistant’s tool implementation should validate input parameters, authenticate through a server-side credential mechanism, apply narrowly scoped authorization, set explicit timeouts, and validate the pricing response before returning it to Claude. Credentials must never be supplied by or exposed to the model. Appropriate logging should capture request attribution, endpoint outcome, and timing without unnecessarily recording sensitive data.

Anthropic describes the Claude API as a RESTful interface and supports custom tools in which the application executes the requested function or API operation. Claude API Overview, Tool Use Overview

MCP could be justified when standardized discovery or reuse across clients is required, but the scenario already supplies a stable API and does not establish that additional requirement.

Study Guide references/topics: Protocol selection; direct API integration; stateless calls; latency minimization; scoped credentials; avoiding unnecessary agentic complexity.

===============

Transaction value is only one component of risk. Even a small refund can create fraud exposure, discriminatory outcomes, contractual violations, account compromise, policy inconsistency, or cumulative financial loss. Delegation should therefore depend on the nature, detectability, reversibility, and aggregate impact of failure—not a monetary threshold alone. Option B supports risk-stratified automation: routine, well-evidenced cases may proceed automatically, while anomalous, identity-sensitive, disputed, or policy-edge cases require review or escalation. Anthropic’s trustworthy-agent principles emphasize maintaining meaningful human control rather than imposing universal review or unrestricted autonomy. Option A is unnecessarily absolute; Option C prioritizes efficiency over governance; and Option D prohibits potentially safe, controlled automation without risk analysis.

Study Guide references/topics: Delegation competency; human control; risk classification; reversibility; cumulative exposure; escalation design.

Per-user, scope-restricted OAuth tokens directly eliminate the shared-credential weakness. Each token represents a distinct user or delegated identity, carries only the permissions required for that user’s role, and can be revoked independently. This supports least privilege, limits the blast radius of credential compromise, and enables authorization decisions to be tied to a specific principal instead of an undifferentiated shared API key.

Actor attribution in tool-call logs directly resolves the second finding. Each audit event should record the authenticated user, delegated service identity, session or request identifier, tool name, authorization decision, relevant resource, timestamp, and outcome. This establishes accountability and permits incident investigators to reconstruct who initiated a consequential action.

Moving credentials into a server-side secret store is sound secret-management practice, but it does not by itself correct the shared-identity problem: a centrally stored key can still represent every user as the same principal. Schema validation protects downstream processing, while retrieval-layer RBAC protects information access; neither addresses the two specific audit findings.

These controls should be reinforced by short-lived tokens, server-side authorization enforcement, sensitive-field redaction, and tamper-resistant audit-log retention.

Study Guide references/topics: [Claude Code security](https://docs.anthropic.com/en/docs/claude-code/security); [MCP authentication and OAuth] (https://docs.anthropic.com/en/docs/claude-code/mcp); least privilege; identity propagation; auditable tool execution.

===============

Project scope is designed for configuration that must be shared by everyone working in a repository. Committing the approved configuration provides versioning, review history, repeatable onboarding, and consistent MCP and permission behavior across the monorepo. User scope is intended primarily for individual preferences and personal rules, so independently maintained user files create configuration drift. Local scope affects only one person in one project and is therefore unsuitable for team-wide distribution. A lead architect’s local configuration cannot establish a shared environment. Claude Code documentation specifically describes .claude/settings.json as the shared project settings file that teams check into source control. Security-critical policies that must not be overridden should additionally be enforced through managed organizational settings rather than relying exclusively on repository configuration. Claude Code: Settings files and precedence

================

Yes, Yes, No, Yes, No

Zero-shot prompting is a reasonable baseline for a clearly defined closed-set classifier because the model receives both the permissible labels and their meanings. Performance should then be measured against an evaluation set before additional prompt complexity is introduced. Few-shot examples are also sound for extraction because examples containing the exact field names demonstrate the required schema and reduce formatting ambiguity. Anthropic identifies examples as one of the most reliable methods for steering output format, structure, and consistency. Prompting Best Practices

Chain-of-thought is not an unconditional default. Additional reasoning is appropriate when the task requires substantive decomposition or complex judgment, but routine classification and extraction may gain little while consuming more tokens and increasing latency. The technique should be selected according to evaluated task requirements.

An even-handed comparison governed by explicit criteria mitigates leading-question bias by requiring symmetrical treatment of alternatives rather than presupposing a preferred conclusion. Finally, prompt validation is model-specific. New model versions may follow instructions differently, change formatting behavior, alter tool selection, or produce different latency and cost characteristics. Anthropic’s migration guidance consequently recommends reevaluating prompts and workloads against the new baseline. Model Migration Guide

Study Guide references/topics: Zero-shot prompting; few-shot examples; reasoning techniques; neutral framing; model migration and regression evaluation.

===============

The sponsor’s statement expresses a desired outcome but does not define an implementable requirement. Option B converts the vague request into concrete information about users, workflow triggers, expected resolutions, decision boundaries, and non-functional constraints. These findings can then be translated into measurable success criteria, evaluation cases, integration requirements, and escalation rules. Anthropic’s evaluation guidance states that an LLM application should begin with specific, measurable, achievable, and relevant success criteria. Option A prematurely treats an aspiration as a complete specification. Option C incorrectly transfers discovery responsibility entirely to the sponsor, while Option D imports assumptions from an unrelated context. Effective Claude architecture requires traceability from the business problem through requirements, evaluation measures, design decisions, and production acceptance criteria. Anthropic: Define success criteria and build evaluations

================

Claude Code permission decisions can be affected by multiple configuration scopes. The appropriate diagnostic action is to inspect the effective rules and identify the exact deny pattern. Managed policy has the highest precedence, and a deny rule from any applicable scope cannot be overridden by an allow rule elsewhere. The architect must then determine whether the denial represents an intentional security boundary or an incorrectly scoped rule. If correction is justified, only the specific rule should be amended. Repeated retries cannot change a deterministic permission decision. Disabling permissions or granting unrestricted Bash access removes safeguards far beyond the reported problem and violates least privilege. Anthropic’s permission documentation explicitly describes scope precedence and deny-first evaluation. Claude Code permissions

Project-scoped configuration is the appropriate mechanism for repository-specific settings that must remain consistent across collaborators. Claude Code recognizes `.claude/settings.json` as the shared project settings file and `.mcp.json` as the project-level MCP server configuration. When these files are committed to source control, every engineer receives the same definitions through the repository’s normal clone, pull, review, and versioning workflow.

Anthropic’s scope hierarchy distinguishes project configuration from user and local configuration. Files under the user scope apply to one engineer and are not automatically shared. Operating-system environment variables are also workstation-specific and require separate configuration management. Managed settings can be distributed centrally, but they are primarily intended for organization-wide security policies and controls that must be enforced by administrators; they are less appropriate when the configuration belongs specifically to one repository.

Shared configuration should contain server definitions and environment-variable references, but not plaintext credentials. Each engineer can resolve required secrets through an approved local secret mechanism while retaining identical server names, transports, arguments, and non-sensitive settings. Project MCP servers also remain subject to workspace trust and approval controls before execution.

Study Guide references/topics: [Claude Code configuration scopes](https://docs.anthropic.com/en/docs/claude-code/settings); project-scoped MCP servers; `.mcp.json`; repository-based operational standardization; secure secret separation.

===============

Option D defines a measurable performance characteristic rather than a business capability. The requirement specifies a percentile-based latency threshold—under 800 milliseconds at the 95th percentile—and ties that threshold to the expected workload volume. It therefore describes how well the system must operate under defined conditions.

Options A, B, and C are functional requirements. They describe actions the solution must perform: extracting fields and creating a downstream record, producing an editable draft, and assigning tickets to categories. Functional requirements define observable system behaviors or outputs.

A strong non-functional requirement must include a metric, threshold, measurement method, and operating condition. For Option D, the architect should further define whether latency is measured end-to-end or only at the model API, the test duration, concurrency, input-size distribution, cache state, timeout behavior, geographic region, and treatment of failed requests. Without those details, teams may report incompatible results while claiming to satisfy the same requirement.

Non-functional requirements for Claude solutions commonly address latency, availability, throughput, cost, privacy, security, recoverability, observability, scalability, and regulatory compliance.

Study Guide references/topics: Structured discovery; functional requirements; non-functional requirements; p95 latency; workload conditions; measurable acceptance criteria.

===============

Least privilege means granting the exact capabilities required for the defined responsibility—no more and no less. Option B permits the assistant to read the approved knowledge base and write only to the draft queue. Enforcing the allow-list at the orchestration layer creates a deterministic authorization boundary outside the model’s prompt and reasoning.

The rule should also be role-specific so different users or operational roles receive only the tools and data scopes they require. The knowledge-base tool should remain read-only, while the draft-queue tool should be restricted to draft creation rather than external sending, approval, or production publication.

Option A exposes every globally readable tool, creating unnecessary access to unrelated information. Option C relies on voluntary model behavior instead of enforceable authorization. A mistaken tool selection or successful prompt injection could therefore invoke an inappropriate capability. Option D minimizes permissions by eliminating the assistant’s required functionality and is not a viable least-privilege implementation.

Anthropic’s permission guidance distinguishes allow, ask, and deny rules and confirms that these controls are enforced by Claude Code rather than by model instructions. Configure Permissions

Study Guide references/topics: Least privilege; per-role authorization; orchestration-layer enforcement; capability scoping; read/write separation; tool allow-lists.

===============

Requirements must first establish what the capability is expected to accomplish, who will use it, and which constraints bound the solution. Option B captures the business purpose, success criteria, and intended user population. Option E identifies critical non-functional requirements that will influence model selection, integration patterns, deployment boundaries, evaluation design, and operational controls. These inputs must exist before stakeholders can validate whether the captured requirements accurately represent their needs. Architecture decision records are produced after meaningful alternatives and design decisions exist. Detailed traceability is important but follows the initial elicitation and consolidation process. Rollout scheduling occurs later, after requirements and architecture have been validated. Anthropic likewise recommends defining measurable success criteria and empirical evaluation methods before optimizing prompts or finalizing the implementation. Anthropic: Prompt engineering overview

================

Approval-rate parity measures whether groups receive positive recommendations at similar rates. It does not establish whether false approvals, false denials, sensitivity, specificity, or calibration are comparable across those groups. A system can therefore satisfy aggregate approval parity while imposing materially different error burdens on protected populations. The evaluation must include subgroup confusion matrices, false-positive and false-negative rates, calibration, intersectional analysis, and confidence intervals.

The 94 percent agreement rate measures fidelity to human underwriter decisions, not fairness. Human decisions are not automatically unbiased ground truth. If historical underwriting practices contain structural, procedural, or measurement bias, a model that reproduces those decisions accurately can reproduce the same bias. The reference labels must therefore be independently assessed for legitimacy, consistency, and potential discriminatory effects.

Nothing in the scenario establishes that protected groups were omitted, so D is unsupported. Similarly, sample size may require examination, but the scenario provides no statistical information proving that sample size is the principal deficiency. The evidence already contains two identifiable conceptual gaps regardless of sample size.

Study Guide references/topics: [Defining multidimensional evaluation criteria](https://docs.anthropic.com/en/docs/build-with-claude/develop-tests); fairness measurement; subgroup error analysis; label and benchmark bias; human-baseline limitations; governance evidence.

===============

Options A and B describe model-specific behavioral risks. An overly broad safety response is a false positive caused by how the language model interprets contextual or surface-level signals. A plausible but unsupported answer is a hallucination: the generated statement appears credible but lacks grounding in the retrieved evidence. Anthropic recommends grounding factual responses in source material, requiring citations, allowing Claude to express uncertainty, and verifying generated claims. Options C, D, and E are conventional infrastructure or application failures involving certificates, routing, and database connectivity. They can occur in systems that do not use an LLM and should be addressed through standard monitoring, configuration management, retry policies, and operational testing. Anthropic: Reduce hallucinations

================

A single confirmed fabrication requires investigation, but it does not establish the population-level failure rate. Sampling recent summaries determines prevalence, affected content types, and whether the incident correlates with particular prompts, documents, or model versions. The architecture must also be examined for grounding controls, including source-constrained generation, citation rendering, claim-to-source verification, and rejection of unsupported statements. Anthropic recommends making factual outputs auditable through citations and requiring supporting quotations for important claims. Immediate universal shutdown may become necessary if the risk is severe, but doing so before establishing scope is not the most disciplined first response. Options D and E dismiss evidence without validation and would prevent corrective action. Reducing hallucinations

A steering committee needs an answer-first recommendation that states the proposed architectural decision, the business outcome it supports, and the approval or action required. After establishing that common decision frame, the document should address each stakeholder’s concerns separately.

The CFO requires implementation and operating costs, expected savings, consumption assumptions, financial exposure, and cost-control mechanisms. General counsel needs evidence concerning confidentiality, privilege, data handling, human legal review, explainability, retention, and regulatory risk. The CIO needs integration architecture, security boundaries, availability, scalability, operational ownership, technical dependencies, and lifecycle support.

Option A begins at an implementation depth unsuitable for the entire committee. Option B overweights the CFO’s concerns and delays the rationale required by the other decision-makers. Option C ignores audience-specific responsibilities and forces every stakeholder to extract relevant information from an undifferentiated narrative.

The recommended structure is: decision and executive rationale, expected outcomes, accepted trade-offs, stakeholder-specific impact sections, major risks and controls, implementation implications, success metrics, and the requested governance decision. Technical appendices can preserve component-level detail without overwhelming the principal recommendation.

Study Guide references/topics: Executive communication; stakeholder differentiation; decision-first structure; financial, legal, and technology concerns; architectural recommendations.

===============

Yes, Yes, No, Yes, No

More capable reasoning configurations generally require greater computation, produce higher latency, and may consume larger token or rate-limit allocations than lighter alternatives. This makes the first statement generally accurate, although actual performance must be measured under the organization’s specific provider and service tier.

Representative workload testing is essential because generic benchmark rankings do not establish performance on the organization’s prompts, data, edge cases, tools, or required output format. Anthropic recommends benchmark tests specific to the use case, using actual prompts and data, followed by comparison of accuracy, quality, edge-case behavior, and cost. Choosing the Right Model

The strongest model is not automatically the correct production choice. Model selection must balance capability, speed, cost, throughput, and operational constraints. Tiered routing implements this principle by serving predictable requests with an efficient model and escalating difficult cases to a more capable model. Anthropic documents multi-model strategies in which lower-cost workers handle most traffic while harder decisions are escalated.

Recency is also not a valid selection criterion. A newly announced model must pass the workload’s existing evaluation, security, latency, and cost gates before replacing a validated production model.

Study Guide references/topics: Model trade-offs; representative evaluations; tiered routing; escalation architectures; cost–capability optimization; model migration validation.

===============

Adversarial tests directly evaluate behavior under attack, including prompt injection, malformed records, malicious document content, credential-extraction attempts, and unauthorized tool requests. Integration tests provide the complementary coverage required by the question: they verify authentication, retrieval, prompt construction, model invocation, validation, logging, and downstream actions as one end-to-end system. Regression, smoke, and unit testing remain valuable, but they do not jointly satisfy the two stated objectives. A stable regression set measures previously established behavior; smoke tests provide shallow availability confirmation; and prompt-rendering unit tests cover only one component. Anthropic recommends explicit prompt-injection defenses and task-specific evaluations that represent both normal and edge-case traffic. Prompt-injection mitigation

The network-boundary requirement is a mandatory architectural constraint; therefore, Option D must be resolved before evaluating operational savings. Managed infrastructure can reduce responsibility for agent loops, sandboxes, session state, recovery, and orchestration, but those advantages are immaterial if financial data leaves the approved boundary. Anthropic describes Claude Managed Agents as a hosted agent harness and provides configurable data-residency controls. The architect must still validate inference location, session persistence, encryption, subprocessors, credential handling, auditability, and deletion behavior against the organization’s exact security policy. Options A, B, and C affect migration effort or economics, but they cannot override a non-negotiable data-governance requirement.

Study Guide references/topics: Architectural constraints; managed versus self-hosted agents; data boundaries; residency; security-gated decision-making.

Option C introduces a deterministic validation boundary between probabilistic model output and downstream execution. Structured validation can reject missing fields, invalid types, unauthorized actions, malformed arguments, and values outside approved enumerations before any tool or business process acts on the response. Anthropic recommends Structured Outputs when guaranteed schema conformance is required. Increase Output Consistency

Option D implements defense in depth. Prompt-level instructions establish intended behavior, while runtime classifiers, content filters, policy engines, output validators, tool permissions, and human confirmation gates provide controls outside the model. Anthropic’s guardrail guidance recommends combining safeguards rather than depending on one mechanism. Mitigate Jailbreaks and Prompt Injections

Option A creates a bypass mechanism that adversaries may discover or imitate. Option B may alter output variability but is not a security boundary and cannot guarantee policy compliance. Option E places complete reliance on a system prompt, which remains vulnerable to ambiguity, conflicting inputs, model errors, and indirect prompt injection.

Guardrails should be risk-based, independently enforceable where possible, observable, tested against adversarial cases, and proportional to the consequences of downstream actions.

Study Guide references/topics: Layered guardrails; structured validation; runtime content checks; prompt injection; deterministic enforcement; defense in depth.

===============

Claude Code constructs the model context before the first user message. That initial context can include the system prompt, project instructions, subagent configuration, MCP server instructions, and tool definitions expressed as JSON schemas. When numerous MCP servers expose many tools and Tool Search is disabled, those schemas must be loaded upfront. A large collection of verbose tool descriptions and parameter schemas can therefore consume tens of thousands of tokens before the engineer enters any text.

Tool Search is specifically designed to prevent this form of context inflation. With Tool Search enabled, Claude initially receives only server instructions and compact tool-identification information; full tool schemas are deferred until a relevant tool is discovered and required. Anthropic consequently describes Tool Search as the mechanism for keeping MCP context consumption low as integrations scale.

Keyboard layouts, font rendering, and display scaling operate outside the language-model context and cannot generate API input tokens. Likewise, model behavior does not silently reserve 50,000 tokens for personal preferences. The diagnostic action is to inspect configured MCP servers and tool counts, determine whether Tool Search has been disabled, and remove or defer unnecessary definitions.

Study Guide references/topics: [MCP Tool Search and context scaling](https://docs.anthropic.com/en/docs/claude-code/mcp); MCP server configuration; subagent context budgeting; tool-schema optimization.

===============

AI-generated code requires verification through the same controlled engineering workflow as human-authored changes. Removing review eliminates the gate that can detect incorrect assumptions, security defects, missing tests, and unintended behavior. Line count is not a defensible risk measure: a one-line authorization change, dependency alteration, or command-execution defect can be more consequential than a large documentation update. Review requirements should instead be determined by affected components, privilege, data sensitivity, reversibility, test coverage, and deployment impact. Expanding capacity may be operationally useful but is not the central diligence objection. Raising the threshold retains the same flawed proxy, while rejecting all AI-generated changes discards legitimate productivity benefits. Anthropic states that users remain responsible for reviewing proposed code and commands. Claude Code security

The strongest investigation combines configuration comparison with controlled replay. Comparing the current system prompt and output-schema settings against the last known good version identifies maintenance-related configuration drift. Relevant differences include removed formatting instructions, altered schema requirements, incompatible parser expectations, changed stop sequences, or a disabled structured-output mechanism.

Replaying known premaintenance requests against the current configuration creates a controlled test set. Inspecting the raw response before parsing then establishes whether malformed JSON originates in model generation or is introduced by streaming assembly, transport handling, post-processing, or the downstream parser. If the raw response is valid but the pipeline rejects it, the fault lies after generation. If the raw response is malformed, the prompt, schema enforcement, stop reason, and request configuration become the primary investigation targets.

Clearing the prompt cache assumes cache corruption without evidence. Changing model tiers introduces another variable and weakens causal attribution. Raising `max_tokens` is appropriate only when raw responses show truncation and the returned stop reason indicates the token limit was reached; it is not the most direct initial test.

Study Guide references/topics: [Increasing output consistency](https://docs.anthropic.com/en/docs/test-and-evaluate/strengthen-guardrails/increase-consistency); [agent evaluation methodology] (https://www.anthropic.com/engineering/demystifying-evals-for-ai-agents); regression replay; last-known-good comparison; raw-output inspection.

===============

Security and compliance reviewers must determine whether material threats have been identified, whether required controls address those threats, how remaining exposure will be accepted, and whether the decision can be demonstrated during an audit. Option C supplies exactly that evidence.

The threat model establishes protected assets, trust boundaries, adversaries, attack paths, and likely failure modes. Control mappings connect technical and procedural safeguards to organizational policies, regulatory requirements, and assurance frameworks. Residual-risk acceptance criteria define which risks may remain after mitigation, who has authority to accept them, and when further treatment is mandatory. Audit traceability connects requirements and decisions to implementation evidence, evaluations, approvals, logs, and operational monitoring.

Option A is designed primarily for product management. Option B addresses engineering planning and execution. Option D is appropriate for an executive sponsor who requires business justification and a concise risk summary but normally does not perform detailed control verification.

Audience adaptation should not alter the architectural decision itself. It changes the depth, terminology, evidence, and ordering used to communicate the decision. Security and compliance reviewers require verifiable control evidence rather than only schedule, implementation, or business-level summaries.

Study Guide references/topics: Audience-specific communication; threat modeling; compliance control mapping; residual-risk acceptance; audit evidence; architectural decision traceability.

===============

The required sequence must reduce repeated computation without changing the information or model capability used to generate projections. Prompt caching the static 6,000-token policy preamble directly addresses the largest cost contributor while preserving the complete instruction set. Anthropic states that cache reads cost substantially less than uncached input tokens, making repeated, stable prompt prefixes an ideal caching target. Prompt Caching

Caching frequently reused historical-sales chunks applies the same principle to the retrieval layer. When identical, version-controlled chunks are repeatedly fetched and supplied to the model, caching their retrieval or reusable representation eliminates redundant work while maintaining the same evidence available to the projection process. The cache must use appropriate invalidation or source-version keys so updated sales data cannot be replaced by stale content.

Options A and B modify the information supplied to Claude. Fewer sales chunks could remove relevant historical evidence, while truncating policy instructions could alter constraints or projection behavior. Option D introduces a model-capability change and therefore cannot guarantee unchanged accuracy without a comparative evaluation. Those interventions may be considered later, but only after representative regression testing establishes equivalence.

Study Guide references/topics: Cost profiling; prompt caching; retrieval caching; cache invalidation; accuracy-preserving optimization; model and context trade-offs.

===============

Regression testing determines whether a change has broken behavior that previously satisfied the defined acceptance criteria. A stable reference set preserves known-good cases and enables the updated prompt to be compared with the prior baseline using consistent scoring rules. This makes Option B directly aligned with the described failure. Adversarial testing focuses on malicious or boundary inputs, such as prompt injection. Smoke testing confirms that the system remains broadly available, but it does not measure whether previously correct model behavior has degraded. Integration testing verifies contracts and interactions among components and may not detect output-quality regressions caused specifically by a prompt change. Anthropic recommends task-specific evaluations that mirror real workloads, include edge cases, and measure performance against explicit success criteria throughout prompt iteration. Anthropic: Define success criteria and build evaluations

================

The absence of company-specific training does not establish unbiased behavior. A foundation model may reflect demographic associations or unequal treatment patterns originating from pretraining data, model behavior, prompt framing, retrieval content, or the surrounding decision workflow. Because candidate screening can materially affect employment opportunities, the complete system must be evaluated across relevant protected groups before deployment. Appropriate measures include selection-rate differences, false-positive and false-negative rates, calibration, intersectional slices, and human override patterns. Disclosure that AI is used does not replace empirical fairness testing. Protected-class labels need not have been explicit in training data for proxy variables or inferred attributes to produce disparities. Provider policies similarly do not validate a customer’s specific implementation. Anthropic system cards document discriminatory-bias evaluations, reinforcing the need for application-level testing. Claude model system card

Haiku is the appropriate starting point because the workload is high-volume, latency-sensitive, and based on a stable closed set of moderation labels. These characteristics favor a fast, cost-efficient model capable of consistent classification without incurring the additional inference time and expense associated with deeper reasoning.

Anthropic’s model-selection guidance requires architects to balance capability, speed, and cost rather than automatically selecting the most capable model. Its content-moderation guidance specifically identifies Haiku as a cost-effective option for processing moderation workloads at substantial scale. Choosing the Right Model, Content Moderation

Opus is disproportionate to a routine closed-set classification problem. Sonnet may become justified if evaluation demonstrates that Haiku fails materially on complex policy distinctions, multilingual ambiguity, or adversarial edge cases, but it should not be selected merely because it is larger. Enabling extended thinking on every request would further increase latency and token consumption without evidence that the additional reasoning improves the defined success metrics. The correct architectural practice is to establish a representative moderation evaluation set, validate Haiku against accuracy and safety thresholds, and escalate only the cases that genuinely need deeper reasoning.

Study Guide references/topics: Model selection; capability–latency–cost trade-offs; classification workloads; evaluation-driven routing; moderation architecture.

===============

Alignment is an ongoing governance process rather than a one-time approval. Option D creates a predictable cadence for reassessing objectives, constraints, assumptions, risks, service levels, and success measures as business and technical conditions change. Option E requires disagreements to be surfaced and reconciled explicitly, preserving decision traceability and preventing unresolved conflicts from reappearing during implementation or operation. The routine should document decisions, owners, accepted trade-offs, action items, and conditions that trigger reconsideration. Recording meetings may improve accessibility but does not itself produce alignment. Rotating facilitation is optional and may weaken accountability. Automatically escalating every disagreement to an executive sponsor bypasses normal architectural negotiation and should be reserved for material deadlocks.

Study Guide references/topics: Stakeholder alignment; feedback cadence; assumption management; conflict reconciliation; decision records; lifecycle governance.

Project scope is designed for repository-specific configuration shared through version control, including MCP server definitions, project permissions, hooks, and team instructions. This gives every contributor a consistent baseline while keeping changes reviewable. Managed configuration is appropriate for mandatory security and compliance controls because administrators can deploy it centrally and prevent lower scopes from overriding it. Local scope is machine-specific and unsuitable for organization-wide enforcement. User scope follows one engineer across projects, so it cannot guarantee that an entire team receives identical MCP definitions. Personal editor preferences belong in user or local configuration, not project scope, unless the team has explicitly standardized them. Anthropic’s settings hierarchy distinguishes managed, project, local, and user responsibilities. Claude Code settings