Anthropic CCA-F (Claude Certified Architect - Foundations (CCAR-F)) Practice Test 34

Question 330

A company wants only one repository to use a database migration skill, not every personal project. Which placement is appropriate?

  1. Store the release workflow as a project skill with clear instructions
  2. Retain supporting resources within the skill and load them when needed
  3. Store the skill at project scope rather than globally for the user
  4. Constrain skill tool permissions to the necessary read-oriented operations

Explanation

Correct Answer

C:

Scope limits where reusable workflows are discovered and available. A migration operation is repository-specific and should not automatically appear in unrelated work.

Incorrect Answers

A:

A reusable skill can collect steps, supporting files and tool permissions under one invocable workflow; its benefits for making a repeatable release workflow reusable do not resolve the need for limiting a sensitive skill to its repository.

B:

Skills can bundle reference material and executable helpers for a focused task; supporting packaging on-demand skill resources cannot replace the required limiting a sensitive skill to its repository.

D:

A skill's allowed tool surface should match its declared job; that serves defining least-privilege tools for a skill, whereas the case requires limiting a sensitive skill to its repository.

Question 331

The team wants a review process that can be repeated after model or prompt updates. Which operational practice supports this?

  1. Choose an independent review pass with source requirements and a separate rubric
  2. Version prompts and rubrics and rerun a fixed regression evaluation suite
  3. Confirm claims against sources before a separate editorial clarity pass
  4. Build in independent evidence checks rather than relying solely on answer agreement

Explanation

Correct Answer

B:

Comparable benchmarks reveal behavior changes when the system evolves. Without versioned tests, anecdotal improvements can mask degraded accuracy.

Incorrect Answers

A:

For improving independence between author and reviewer, independent evaluation can challenge design assumptions rather than merely reaffirm the author; however, this requirement concerns making multipass quality evaluations reproducible.

C:

Fact checking and presentation optimization use different evidence and criteria; this may support separating factual checking from writing polish but cannot alone meet the requirement for making multipass quality evaluations reproducible.

D:

Several models may share the same misconception; the stated problem requires making multipass quality evaluations reproducible rather than avoiding false confidence from reviewer agreement.

Question 332

A developer wants to locate definitions of a symbol anywhere in source text before reading specific implementations. Which workflow is efficient?

  1. Grep for the symbol and then Read the matching files
  2. Adopt Glob for pathname discovery before reading content
  3. Select Edit for a targeted modification to an existing file
  4. Search paths with Glob the matching migration paths and then inspect them in order
  5. Carry out the formatter through Bash and inspect the resulting diff or file changes

Explanation

Correct Answer

A:

Searching content narrows candidates; reading the relevant paths supplies surrounding context. Reversing the order can waste context on many unrelated files.

Incorrect Answers

B:

Path globbing matches file names without requiring content search; this addresses searching by path rather than file contents, while the question is about moving from symbol search to targeted file inspection.

C:

Targeted editing changes selected text while maintaining the surrounding source; that relates to changing a limited region of existing code, not the necessary moving from symbol search to targeted file inspection.

D:

Although filename enumeration establishes the migration set without executing anything, the choice addresses gathering migration files without executing code and not the requested moving from symbol search to targeted file inspection.

E:

Process output alone may not prove exactly what changed; the deciding condition is moving from symbol search to targeted file inspection, whereas this action concerns verifying changes produced by an executable command.

Question 333

The same agent can call both publish_report and verify_citations, potentially publishing first. What access design strengthens safety?

  1. Have a privileged diagnostics component inspect logs and return sanitized findings
  2. Enable the model to answer without an obligatory tool invocation
  3. Begin by setting per-task call budgets and terminate or escalate on exhaustion
  4. Gate publish authority on validated workflow state instead of mere tool availability

Explanation

Correct Answer

D:

Tool existence is not business authorization for every stage. State-aware policy prevents premature publication even if the agent asks for the mutation.

Incorrect Answers

A:

Dedicated diagnostics isolates sensitive infrastructure data while providing actionable summaries; segregating diagnostic access from regular agents and making a sensitive tool contingent on workflow state require different controls in this situation.

B:

Optional tool selection prevents unnecessary calls when the request requires no external data; the unsatisfied requirement is making a sensitive tool contingent on workflow state, not avoiding forced calls on tool-free requests.

C:

Tool budgets prevent unbounded external work and let the system deliver a controlled incomplete status; in this scenario, controlling tool invocation cost per task is secondary to making a sensitive tool contingent on workflow state.

Question 334

A tool takes arbitrary SQL as an argument even though the agent only needs order status. What safer interface would you design?

  1. Define explicit amount units and a required currency input
  2. Expose a narrow parameterized order-status operation instead of general SQL
  3. Specify the target resource, permitted query scope and result semantics
  4. Establish typed required parameters with separate identifier semantics
  5. Declare the mutation and external communication effect explicitly in the tool description

Explanation

Correct Answer

B:

A purpose-limited interface minimizes injection and unauthorized data access. The agent should express business intent without acquiring unrestricted query execution.

Incorrect Answers

A:

Financial tools need unambiguous units and typed fields; that addresses disambiguating monetary units in tool calls, whereas this case calls for constraining a tool to least-privilege business operations.

C:

Effective tool descriptions explain when the operation should be used and what it returns; exposing the resource scope of retrieval tools and constraining a tool to least-privilege business operations are distinct responsibilities under these constraints.

D:

A precise input schema can distinguish account identifiers from email addresses and reject unsupported shapes; the action supports distinguishing identity fields in a tool interface but falls short of constraining a tool to least-privilege business operations in this case.

E:

Callers need to distinguish observation from state change before choosing tools; this can serve signaling irreversible tool side effects, but leaves the requirement for constraining a tool to least-privilege business operations unmet.

Question 335

A decision prompt asks for the 'best' architecture with no constraints, producing inconsistent trade-offs. What information should be supplied?

  1. List retryable failures, mandatory escalation conditions and forbidden fallbacks
  2. State evaluation priorities such as latency, correctness, budget and operational ownership
  3. Detail required positive findings, limitations, risks and outstanding decisions
  4. Insist on tool-confirmed completion before wording any definitive customer promise

Explanation

Correct Answer

B:

Architectural choices depend on goals and trade-offs. Explicit priorities make it possible to explain why one design wins over another for this workload.

Incorrect Answers

A:

Different error classes require different branches; this contributes to defining precise failure-handling criteria without meeting the need for making architecture trade-off priorities explicit.

C:

Because balanced section requirements make omissions more visible during validation, the measure supports eliciting both benefits and risks in reports rather than the needed making architecture trade-off priorities explicit.

D:

The prompt can distinguish proposed, attempted and confirmed actions; its benefits for preventing premature promises of side effects do not resolve the need for making architecture trade-off priorities explicit.

Question 336

A support agent mixes a known shipping status with a guessed delivery date and presents both as confirmed. What output discipline helps?

  1. Retain retrieval time and a source version or approved snapshot reference
  2. Maintain an auditable decision record linking tool observations, rules and approvals
  3. Label confirmed observations separately from estimates and unknowns
  4. Classify retrieved document instructions as untrusted content rather than authority

Explanation

Correct Answer

C:

Provenance distinguishes system-of-record data from inference. Clear labels let users understand which details are operational facts and which may change.

Incorrect Answers

A:

Live pages can change or disappear; this may support preserving time-sensitive research provenance but cannot alone meet the requirement for separating verified status from estimates.

B:

Reproducibility requires more than the final recommendation; the stated problem requires separating verified status from estimates rather than reconstructing the evidence behind past decisions.

D:

For resisting prompt injection embedded in source material, external evidence may contain prompt injection; however, this requirement concerns separating verified status from estimates.

Question 337

A research parent needs to combine citations from several subagents. What must each subagent return besides narrative?

  1. Target both delegated instructions and available data connectors to the assigned case
  2. Add locale and other required business constraints in each relevant worker brief
  3. Return stable source identifiers alongside evidence extracts and claims
  4. Specify the need for evidence of which documents and versions the worker actually consulted

Explanation

Correct Answer

C:

Citation-safe merge requires referential metadata. Bare prose cannot reliably distinguish original sources from intermediary agent conclusions.

Incorrect Answers

A:

Context minimization and access controls complement each other; that relates to narrowing a worker's data-access boundary, not the necessary preserving source pointers through delegation.

B:

Although constraints do not reliably propagate by implication across context boundaries, the choice addresses propagating shared constraints into tasks and not the requested preserving source pointers through delegation.

D:

Source and version metadata distinguish inspected facts from memory-based guesses; the option would serve validating the evidentiary basis of a subagent answer, not preserving source pointers through delegation as specified in the stem.

Question 338

A worker fails but the overall process displays 100% progress. Which state metric should be redesigned?

  1. Maintain a partial workflow outcome and identify the unavailable branch explicitly
  2. Categorize schema mismatches as repairable-input defects rather than transient outages
  3. Report completed, failed and pending work as separate outcome categories
  4. Log the evidence gap and either retry safely or abstain from the unsupported claim

Explanation

Correct Answer

C:

Progress counters must reflect task outcomes rather than merely finished attempts. Clear status reporting prevents operators from mistaking errors for completed deliverables.

Incorrect Answers

A:

Partial success is not whole-task success; in this scenario, preserving partial-failure status across worker aggregation is secondary to distinguishing task completion from task success.

B:

When considering stopping futile retries of invalid worker payloads, schema failures typically persist until the contract or payload changes; the requirement for distinguishing task completion from task success remains unmet.

D:

Failed retrieval does not authorize fabricating authoritative state; preventing fabricated facts after upstream errors and distinguishing task completion from task success require different controls in this situation.

Question 339

A pipeline wants to parse review results rather than scrape decorative terminal text. Which output control helps?

  1. Adopt least-privilege credentials and sandbox the untrusted review workload
  2. Request machine-readable JSON output and validate its schema
  3. Generate proposed changes in an isolated workspace or pull request for review
  4. Formalize explicit severity criteria and require evidence for blocking findings

Explanation

Correct Answer

B:

Structured output is easier for CI to parse and gate deterministically. Human-oriented prose and terminal formatting make automated decision logic fragile.

Incorrect Answers

A:

PR diffs can contain prompt injection or malicious scripts; limiting trust for untrusted pull-request code and parsing review output safely in CI are distinct responsibilities under these constraints.

C:

Generated tests may be incorrect or brittle; this can serve isolating generated CI code changes, but leaves the requirement for parsing review output safely in CI unmet.

D:

Reviewer outputs should distinguish verified blockers from optional recommendations; that addresses calibrating automated code-review severity, whereas this case calls for parsing review output safely in CI.

Question 340

A team wants every review comment to point to a real code location. Which output requirement supports downstream publishing?

  1. Demand verifiable evidence before elevating the claim to a blocking finding
  2. Confine allowed tools and repository credentials to read-only review operations
  3. Require file path, line range and finding rationale in a validated result schema
  4. Focus on requesting machine-readable JSON output and validate its schema

Explanation

Correct Answer

C:

Location-aware structured findings allow a bot to attach actionable comments. Narrative-only criticism lacks reliable anchors for automated PR feedback.

Incorrect Answers

A:

Because model assertions are hypotheses until supported by inspected code or tests, the measure supports verifying CI review claims before blocking rather than the needed generating location-specific PR findings.

B:

CI runners often have elevated tokens; its benefits for enforcing read-only CI review permissions do not resolve the need for generating location-specific PR findings.

D:

Structured output is easier for CI to parse and gate deterministically; this contributes to parsing review output safely in CI without meeting the need for generating location-specific PR findings.

Claude Certified Architect - Foundations (CCAR-F) (CCA-F) Practice Tests