Anthropic CCA-F (Claude Certified Architect - Foundations (CCAR-F)) Practice Test 28

Question 270

Claude Code edited a file and also changed a remote database through a tool. A user rewinds the local code checkpoint. What should the agent avoid assuming?

  1. That the remaining remote change should be reported when relevant.
  2. That rewinding local files automatically reversed the remote database mutation.
  3. That the local file state may have been restored according to the selected rewind operation.
  4. That the remote database can be queried to determine its current state.

Explanation

Correct Answer

B:

External side effects need their own verified reversal or reconciliation mechanism. Checkpointing concerns the state it actually tracks. A remote database needs its own authoritative readback and, if appropriate, an authorized compensating operation.

Incorrect Answers

A:

A material external change is part of the user's actual task state.

C:

Restoring tracked local edits is within checkpointing's intended scope.

D:

An authorized readback is an appropriate way to inspect the external effect.

Question 271

After an automated formatting pass, a generated patch also changes a comparison operator. What should the reviewer do?

  1. Approve the operator change because it appears near whitespace edits.
  2. Separate the behavioral change from formatting and verify whether it is intended.
  3. Run the formatter again and treat unchanged formatting output as evidence that the new comparison is correct.
  4. Update expected test outputs to match the changed comparison without checking the original requirement.

Explanation

Correct Answer

B:

Inspecting semantic differences prevents a functional change from being hidden in cosmetic work. The comparison operator can affect which inputs pass a condition even if most of the diff is whitespace. That change needs its own justification and relevant behavioral check.

Incorrect Answers

A:

Proximity to formatting does not establish correctness or authorization.

C:

Formatter idempotence checks formatting stability, not whether a changed operator implements the intended behavior.

D:

New baselines can make tests pass while accepting an unintended semantic change. The operator must first be justified against the required behavior.

Question 272

After Claude Code changes a generator's configuration, CI restores an old generated artifact from cache. The cache key hashes the source files but excludes the generator configuration and version. A clean build produces the correct artifact. What is the durable correction?

  1. Delete the cache once and keep the existing key for later runs.
  2. Include all output-determining inputs in the cache key, including generator configuration and version.
  3. Increase cache retention so the same artifact is reused more consistently.
  4. Edit the restored generated file manually after every cache hit.

Explanation

Correct Answer

B:

A cache hit is valid only when the inputs relevant to the output match. The clean-build result isolates the defect to reusing an artifact under an incomplete identity.

Incorrect Answers

A:

This can unblock one build, but a later configuration-only change can collide with the same incomplete key. The invalidation rule remains wrong.

C:

Longer retention preserves the stale entry; it does not make that entry appropriate for a different generator configuration.

D:

Manual patching bypasses the generator and can diverge from clean builds. It treats stale output as an editing task instead of repairing how the cache identifies its inputs.

Question 273

A request says 'support empty input without changing nonempty behavior.' Claude Code makes empty input work. What else must verification cover?

  1. Only whether the file name stayed unchanged.
  2. Whether the final response contains the word support.
  3. Only whether the new empty-input test passes repeatedly on the same fixture.
  4. Representative nonempty cases to check the explicitly required behavior preservation.

Explanation

Correct Answer

D:

The user's acceptance condition includes both the new empty-input handling and the existing nonempty behavior. Representative existing cases provide evidence against a regression caused by the new branch. Passing the empty case alone satisfies only one half of the stated requirement.

Incorrect Answers

A:

A stable file name gives no evidence about the preserved input behavior.

B:

Response wording does not test either part of the acceptance condition.

C:

Repeated coverage of the new case does not verify preservation of the explicitly required nonempty behavior.

Question 274

Claude Code finds a documentation example using a newer API than the project's pinned dependency. What should it do before implementing the example?

  1. Invent an adapter method with the same name and assume it exists.
  2. Use the newest example unchanged because web pages are always compatible.
  3. Check the pinned version's supported interface or explicitly plan a compatible upgrade.
  4. Remove the version pin without testing.

Explanation

Correct Answer

C:

Version-aware implementation avoids importing behavior the repository cannot execute. Pinned versions define the interface the project can actually import. A necessary upgrade should be a deliberate dependency change with compatibility validation rather than an implicit assumption.

Incorrect Answers

A:

An imagined method does not provide a working compatibility layer.

B:

Current documentation may describe features absent from the installed dependency version.

D:

Uncontrolled upgrades can introduce unrelated compatibility changes.

Question 275

After a UI change, snapshot tests fail. Claude Code can update every expected snapshot automatically. What should it establish before accepting those new baselines?

  1. Which visual differences are intended and whether the changed snapshots accurately represent the required behavior.
  2. That a successful update command guarantees the UI meets all accessibility and behavior requirements.
  3. That every failing snapshot is necessarily obsolete because source code changed.
  4. That updating more snapshots always provides stronger regression coverage.

Explanation

Correct Answer

A:

A snapshot update makes the new output the reference; it does not prove the output is correct. Reviewing the actual differences prevents a visual regression from being accepted merely to make the suite pass.

Incorrect Answers

B:

Writing new reference files does not validate requirements beyond the content actually reviewed and checked.

C:

A source change can introduce an unintended regression, so failure is not by itself evidence that the reference should change.

D:

Baseline quantity does not establish that their expected content is correct or relevant to the requested change.

Question 276

A build creates temporary files with a known task-specific prefix. What is the appropriate cleanup strategy?

  1. Remove only verified task-owned temporary outputs using a narrowly scoped path selection.
  2. Delete the user's entire cache directory because some build files may be there.
  3. Remove all files modified today.
  4. Delete every untracked file in the repository without inspection.

Explanation

Correct Answer

A:

Constrained cleanup avoids deleting unrelated files while removing artifacts created by this task. Untracked and recently modified files are not automatically disposable. Cleanup needs evidence of task ownership and a narrow path set, especially in a shared working tree.

Incorrect Answers

B:

A broad cache deletion exceeds the evidence about which files belong to this task.

C:

Modification date does not identify ownership or whether a file is disposable.

D:

Untracked files may include the developer's work and are not necessarily temporary outputs.

Question 277

Claude Code is authorized to create a commit for its fix. What should it check immediately before staging?

  1. The final diff and intended file set, including accidental secrets or unrelated changes.
  2. Only whether git status reports any modified files.
  3. Whether all untracked files can be added with a blanket command.
  4. Whether the commit message is longer than the patch.

Explanation

Correct Answer

A:

A final content check makes the authorized commit specific to the reviewed fix. Staging should include the intended patch rather than every file visible in status. Inspecting content also catches accidental credentials or build noise before they enter history.

Incorrect Answers

B:

A status summary identifies changed paths but does not reveal whether their contents belong in the authorized commit.

C:

Blind staging can include unrelated user files, secrets, or generated noise.

D:

Message length is not a substitute for checking the commit contents.

Question 278

A prompt asks Claude to write a 'high-quality incident summary,' but reviewers disagree about quality. Which change best makes the task testable?

  1. Specify formal language while leaving coverage and evidence requirements undefined.
  2. Allow reviewers to invent a different definition of quality for every result.
  3. Require the output to describe itself as high quality.
  4. Specify required incident facts, evidence handling, audience, and a measurable length constraint.

Explanation

Correct Answer

D:

Observable requirements let reviewers evaluate the same properties instead of relying on an undefined quality label. The criteria should identify what a successful incident summary enables the reader to understand or do. Reviewers can then score coverage and support separately from stylistic preferences.

Incorrect Answers

A:

A style rule still leaves substantive success criteria open to incompatible interpretations.

B:

Changing criteria across reviews prevents consistent measurement of the intended summary quality.

C:

Self-description is not evidence that the summary meets any substantive requirement.

Question 279

A summarization prompt says 'Use only the supplied document' and later says 'Fill every missing fact from general knowledge.' What should the designer do?

  1. Keep the conflicting rules but require every completed field to have the same confident tone.
  2. Place the source-only rule earlier and assume later contradictory wording will never affect the result.
  3. Resolve the conflict with an explicit missing-evidence policy consistent with the task.
  4. Keep both instructions and let each run choose its own interpretation.

Explanation

Correct Answer

C:

A single coherent evidence rule tells the model how to handle absent information. If external enrichment is allowed, it should be explicitly separated from document extraction and attributed. Otherwise missing facts remain missing rather than being silently supplied from model knowledge.

Incorrect Answers

A:

Tone cannot resolve whether unstated facts may be added. It may instead conceal the difference between source evidence and unsupported completion.

B:

Reordering leaves both instructions active. An explicit missing-evidence policy is needed to remove the contradiction rather than hope one placement wins.

D:

Unresolved contradictory instructions make factual boundaries unpredictable across runs.

Question 280

An engineering postmortem must be summarized for customer-support staff. Which prompt detail most directly improves usefulness?

  1. Ask for the longest possible technical derivation.
  2. Use a generic instruction to sound impressive.
  3. Specify audience knowledge, customer impact, workarounds, and escalation needs.
  4. Preserve every internal acronym without definition.

Explanation

Correct Answer

C:

Audience and action needs guide which technical facts to retain and how to explain them. Support staff need information that changes how they assist customers. Internal implementation details belong only where they explain impact, workarounds, or escalation decisions for that audience.

Incorrect Answers

A:

Maximum technical depth may obscure the practical information support staff need.

B:

An impression-focused instruction does not identify the information needed by support staff.

D:

Unexplained terminology can make the summary less accessible to its intended readers.

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