Anthropic CCA-F (Claude Certified Architect - Foundations (CCAR-F)) Practice Test 27

Question 260

A code review automation runs Claude Code on inbound pull requests. An agent is asked to review a feature without knowing its acceptance criteria. Which implementation decision is the best fit?

  1. Switch from review to authorized bounded editing and validation
  2. Run allowed checks and inspect logs before proposing edits
  3. Use an interview or planning step to collect expected behavior
  4. Run read-only inspection and feedback generation without write-capable tools
  5. Inspect tainted inputs, query construction, and existing parameterization

Explanation

Correct Answer

C:

Undefined success conditions make a code change hard to test; targeted questions can establish the missing contract.

Incorrect Answers

A:

Separation between analysis and implementation gives the maintainer control over when automated modifications occur; the alternative would make sense if a human explicitly requests a patch after reviewing generated suggestions. It does not directly fix the observed condition: an agent is asked to review a feature without knowing its acceptance criteria.

B:

When a test suite is failing and the team wants diagnosis only, this would be useful because diagnostic evidence is needed to locate the cause; granting automatic write access in the discovery phase increases change risk; this item instead calls for a way to clarify requirements before implementation.

D:

A CI reviewer should not silently modify a contributor's branch while supposedly assessing it; the relevant precondition would be the pipeline must analyze a PR without altering its contents. The question instead describes an agent is asked to review a feature without knowing its acceptance criteria.

E:

This measure fits a different trigger: a reviewer finds a potential SQL-injection path in an unfamiliar module. In that setting, Security findings require traceable evidence; a pattern match alone may yield a false positive or overlook mitigations; it does not resolve the present need to clarify requirements before implementation.

Question 261

A support agent returns structured case-resolution decisions. The response schema accepts any account ID string, including empty values. What should the architect specify in the design?

  1. Include conditional reason fields tied to the decision status
  2. Define typed decision, rationale, and policy-reference fields
  3. Validate nonempty identifiers and documented format requirements
  4. Run policy and permission checks independently of JSON shape

Explanation

Correct Answer

C:

Field-level constraints catch malformed records before downstream calls, although existence and authorization still require service checks.

Incorrect Answers

A:

A yes/no value alone is often insufficient for safe escalation or customer communication when the answer is no; the alternative would make sense if the model returns a Boolean eligibility flag but omits the reason for ineligibility. It does not directly fix the observed condition: the response schema accepts any account ID string, including empty values.

B:

A consistent structured record supports auditability and guards against free-form messages that omit why the decision was made; the relevant precondition would be a refund decision needs a rationale plus a reviewed policy reference. The question instead describes the response schema accepts any account ID string, including empty values.

D:

When a decision object passes schema validation but requests an unauthorized refund, this would be useful because structured output can be syntactically perfect while semantically unsafe; the execution boundary must reject unauthorized actions; this item instead calls for a way to add local syntactic constraints.

Question 262

A review agent must obey application-level guarantees under adversarial prompts. A tool can modify thousands of records although the task requires one record. Which change most directly addresses this condition?

  1. Validate target identifier and cardinality in tool arguments before execution
  2. Apply a confirmation checkpoint enforced by the host workflow
  3. Persist the rejection and close the pending mutation authorization
  4. Check a central rate or quota counter at execution time

Explanation

Correct Answer

A:

Target and batch-size validation blocks unintended bulk changes even when a model builds syntactically valid arguments.

Incorrect Answers

B:

A required approval checkpoint converts a human-control requirement into execution logic, rather than merely advising the model to ask first; the alternative would make sense if a user requests deletion, but the policy requires independent confirmation. It does not directly fix the observed condition: a tool can modify thousands of records although the task requires one record.

C:

This measure fits a different trigger: a human reviewer rejects a proposed account adjustment. In that setting, A durable rejection should invalidate the action token; otherwise a replayed model output might still reach the mutation endpoint; it does not resolve the present need to constrain the operation to intended resource scope.

D:

When a tool operation must obey a daily account limit, this would be useful because per-prompt reminders cannot coordinate usage across multiple agents; an authoritative service counter can enforce shared limits; this item instead calls for a way to constrain the operation to intended resource scope.

Question 263

A research pipeline emits structured literature reviews. A summarized study reports a percentage without a denominator. What is the most defensible next step?

  1. Check the identifier format and request a corrected value when supported by the source
  2. Require the base population or flag the statistic as incomplete
  3. Record initial failures, corrected fields, and validator outcomes
  4. Check claims against retrieved sources and reject unsupported assertions

Explanation

Correct Answer

B:

A number without its denominator can be misleading; semantic validation must check whether the claim is interpretable.

Incorrect Answers

A:

A targeted citation check catches formatting mistakes before publication while keeping evidence matching as a separate step; the relevant precondition would be a citation record has a malformed DOI string. The question instead describes a summarized study reports a percentage without a denominator.

C:

When a publisher needs to know which assertions were repaired automatically, this would be useful because a repair trail makes changes auditable and helps identify extraction defects hidden by a final clean object; this item instead calls for a way to enforce complete quantitative evidence.

D:

This measure fits a different trigger: the model keeps producing unsupported claims despite clean JSON. In that setting, Schema enforcement is not fact checking; a reviewer must compare meaning with evidence rather than regenerate well-formed guesses; it does not resolve the present need to enforce complete quantitative evidence.

Question 264

A company processes thousands of independent documents through Claude. Thousands of batch results must be matched back to source documents. Which approach satisfies the stated constraint?

  1. Record per-request success or error and retry only eligible failures
  2. Submit independent requests as a batch and reconcile completed results later
  3. Use a normal request-response call for the interactive decision
  4. Attach stable custom identifiers and reconcile outputs with the submission manifest

Explanation

Correct Answer

D:

Result ordering alone may not match input ordering; explicit IDs prevent extracted data being assigned to the wrong document.

Incorrect Answers

A:

This measure fits a different trigger: one batch item fails while the remaining documents complete. In that setting, A batch is not necessarily all-or-nothing; item-level bookkeeping avoids rerunning already successful documents; it does not resolve the present need to preserve request-to-document correlation.

B:

Batch execution suits large offline workloads where delayed completion is acceptable and each request has its own input and output identity; the relevant precondition would be each document is self-contained and does not require an immediate interactive response. The question instead describes thousands of batch results must be matched back to source documents.

C:

Deferred batch completion is a poor fit when a customer is waiting for an immediate answer; workload latency requirements control the choice; the alternative would make sense if a support agent needs a reply within the current live conversation. It does not directly fix the observed condition: thousands of batch results must be matched back to source documents.

Question 265

A research publishing workflow uses confidence-aware human oversight. A model reports ninety-nine percent confidence without a calibrated evaluation process. Which mechanism should be selected?

  1. Apply the checked fix without adding a human approval queue
  2. Require a verifiable source or explicit review of the unsupported claim
  3. Flag both claims and their evidence for accountable adjudication
  4. Analyze error patterns and update targeted tests or escalation criteria
  5. Use observed error rates or rubric-based review rather than the unsupported percentage

Explanation

Correct Answer

E:

Language-model confidence expressions are not calibrated statistical guarantees unless validated against representative outcomes.

Incorrect Answers

A:

Review resources can focus on uncertain decisions while deterministic cosmetic changes proceed under normal controls; the alternative would make sense if a low-risk spelling correction is supported by deterministic lint feedback. It does not directly fix the observed condition: a model reports ninety-nine percent confidence without a calibrated evaluation process.

B:

This measure fits a different trigger: an automated report cites a source that cannot be retrieved by the reviewer. In that setting, Publication quality depends on accessible supporting material, not merely on a plausible citation string; it does not resolve the present need to avoid treating raw self-confidence as probability.

C:

Conflicting high-impact evidence warrants review even if a model prefers one narrative with high confidence; the relevant precondition would be the source evidence is contradictory and would change an important recommendation. The question instead describes a model reports ninety-nine percent confidence without a calibrated evaluation process.

D:

When a human reviewer frequently overrides one category of model decisions, this would be useful because systematic override patterns reveal where the model's decision boundary is weak and where review effort is most valuable; this item instead calls for a way to avoid treating raw self-confidence as probability.

Question 266

An engineering organization maintains Claude Code commands and Skills. A command can deploy production, but a user can trigger it easily. What should be implemented at the appropriate boundary?

  1. Make the long routine explicitly invocable rather than persistent root memory
  2. Require actual deployment permission and approval outside the command text
  3. Keep the project's command or skill files under repository review
  4. Update references and run a small invocation test

Explanation

Correct Answer

B:

A reusable command is an interface, not a security control; the deployment system must enforce who may perform the action.

Incorrect Answers

A:

When a custom command contains a long fixed prompt that every task automatically loads, this would be useful because optional procedures belong in reusable commands or Skills; always-loaded instructions should stay concise and broadly relevant; this item instead calls for a way to pair convenience with external authorization.

C:

A reviewed definition lets the team examine behavior changes and use the same release routine across contributors; the relevant precondition would be a project-specific release command should be reviewed with code changes. The question instead describes a command can deploy production, but a user can trigger it easily.

D:

This measure fits a different trigger: a Skill refers to resources moved during a repository refactor. In that setting, Reusable workflows depend on accessible files; broken references cause failures even if the top-level description is correct; it does not resolve the present need to pair convenience with external authorization.

Question 267

An architecture review uses Claude subagents for isolated design alternatives. A specialist needs only the deployment directory, not the full monorepo. Which option best meets the operational need?

  1. Record the branch identifier, assumptions, and tested outcomes for each alternative
  2. Fork the session or task state before exploring alternate designs
  3. Revalidate changed files and environment state before relying on resumed context
  4. Give the worker the deployment paths, relevant constraints, and safe tools

Explanation

Correct Answer

D:

Scoping protects context budget and keeps the task anchored to the part of the repository it can actually assess. For the stated case, the check is whether the system can supply minimal scoped working context; this approach supplies that control.

Incorrect Answers

A:

This measure fits a different trigger: the team needs to reproduce why two design branches diverged. In that setting, Explicit branch history makes comparisons auditable; merged prose can otherwise conceal which experiment produced each result; it does not resolve the present need to supply minimal scoped working context.

B:

Branches can evaluate alternatives without contaminating each other's assumptions, preserving a meaningful comparison; the relevant precondition would be architects want to compare two approaches from the same initial repository state. The question instead describes a specialist needs only the deployment directory, not the full monorepo.

C:

A saved transcript is not a guarantee that disk or remote dependencies are unchanged; material state must be checked again; the alternative would make sense if an agent resumes after its local repository changed since the last session. It does not directly fix the observed condition: a specialist needs only the deployment directory, not the full monorepo.

Question 268

A coordinator divides a codebase-migration assessment among specialist agents. A security worker discovers an immediate blocker during parallel migration analysis. What is the most appropriate design choice?

  1. Have the coordinator evaluate worker evidence and generate the consolidated recommendation
  2. Define distinct tasks and structured outputs for each worker
  3. Return a structured blocker status with severity and supporting evidence
  4. Wait for both prerequisite worker results before launching the integration-check task

Explanation

Correct Answer

C:

A typed blocker can halt or reroute downstream steps; a buried sentence is easy for a later summarizer to miss. For the stated case, the check is whether the system can signal a blocking finding without hiding it in prose; this approach supplies that control.

Incorrect Answers

A:

Specialists provide evidence and domain judgments; synthesis ownership prevents conflicting independently published conclusions; the alternative would make sense if the team wants a single conclusion after three worker assessments. It does not directly fix the observed condition: a security worker discovers an immediate blocker during parallel migration analysis.

B:

Explicit outputs make the coordinator's merge traceable and prevent the workers from competing over the same incomplete artifact; the relevant precondition would be one worker is responsible for dependency scanning and another for test-risk analysis. The question instead describes a security worker discovers an immediate blocker during parallel migration analysis.

D:

When an integration test needs both migration and dependency analyses to finish, this would be useful because a join condition prevents testing a proposed change against an incomplete dependency or migration plan; this item instead calls for a way to signal a blocking finding without hiding it in prose.

Question 269

A research writing assistant must deliver claims that can be independently checked. The writer may use only materials approved by the research team. Which response is most reliable?

  1. Require numeric support or qualified sourcing for comparative claims
  2. Require verification of critical numbers before publishing, even if additional review is needed
  3. Require effective dates and labels for superseded rules
  4. List approved source identifiers and prohibit unsupported external assertions

Explanation

Correct Answer

D:

A bounded source contract limits answer drift and gives reviewers a concrete basis for rejecting unsupported material.

Incorrect Answers

A:

A clear evaluation rule distinguishes evidence-based comparison from untestable promotional language; the relevant precondition would be a generated comparison uses qualitative adjectives with no supporting figures. The question instead describes the writer may use only materials approved by the research team.

B:

When two stakeholders prioritize accuracy over speed for high-impact figures, this would be useful because a prompt can express the quality trade-off explicitly; rushing a confident result is inappropriate when a wrong figure has high cost; this item instead calls for a way to define the permitted evidence set.

C:

This measure fits a different trigger: a report must distinguish current policies from historical descriptions. In that setting, Time qualifiers are part of factual correctness when documents describe changing systems; bare claims can mislead; it does not resolve the present need to define the permitted evidence set.

Question 270

A multi-source intelligence assistant must preserve provenance in its conclusions. An OCR system extracts text from a scanned image with uncertain characters. What should the team do to resolve this particular gap?

  1. Record the direct document locator and relevant passage reference
  2. Label the date unknown and qualify conclusions that depend on recency
  3. Record file type, version, denominator, and calculation method for each source
  4. Store each value alongside publication and observation dates
  5. Mark the source as OCR-derived and retain the image or page locator

Explanation

Correct Answer

E:

Image-derived text can contain recognition errors; noting origin supports appropriate human verification. For the stated case, the check is whether the system can preserve content modality and extraction status; this approach supplies that control.

Incorrect Answers

A:

A generic homepage does not let reviewers trace the asserted evidence; precise provenance makes claims auditable; the alternative would make sense if a summary cites the homepage of a website but not the page containing a key claim. It does not directly fix the observed condition: an OCR system extracts text from a scanned image with uncertain characters.

B:

When a retrieved document lacks a reliable publication date, this would be useful because a missing timestamp cannot be repaired by inference from a filename or crawl date; time-sensitive claims need verified dates; this item instead calls for a way to preserve content modality and extraction status.

C:

This measure fits a different trigger: a CSV report and a narrative article disagree about a count. In that setting, A discrepancy may arise from different populations or transformations; provenance details are needed before choosing a value; it does not resolve the present need to preserve content modality and extraction status.

D:

Different dates may describe genuine changes rather than factual contradiction; temporal provenance enables accurate interpretation; the relevant precondition would be two sources give different values for the same metric on different dates. The question instead describes an OCR system extracts text from a scanned image with uncertain characters.

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