Anthropic CCAR-P (Claude Certified Architect - Professional) Practice Test 39

Question 380

A developer's assistant silently updates lockfiles and deploy scripts while solving a documentation typo. Which next step is most defensible?

  1. Remove secrets from prompt files and inject short-lived credentials through a vault-controlled runner.
  2. Collect redacted command, diff, test and approval traces with correlation identifiers.
  3. Configure targeted repository navigation and context selection rules.
  4. Require diff review and restrict edit scope before sensitive file changes.

Explanation

Correct Answer

D:

Human-visible change boundaries catch collateral modifications that a model may introduce. Verify controlled edit permissions for sensitive files with actual repository access and CI controls; host-side controls are needed to guarantee controlled edit permissions for sensitive files.

Incorrect Answers

A:

This is a valid architectural concern when optimizing secure credentials for ai-assisted development. Here it leaves controlled edit permissions for sensitive files unaddressed, whereas the intended change must require diff review and restrict edit scope before sensitive file changes before the case can be considered resolved.

B:

This would make sense as an additional project activity for auditable coding-agent activity. The priority required here is controlled edit permissions for sensitive files, so the workflow still needs a control that will require diff review and restrict edit scope before sensitive file changes.

C:

By itself, this improves repository-aware context selection, but the case is not asking the architect to optimize that dimension. The requirement is controlled edit permissions for sensitive files, supported by an approach to require diff review and restrict edit scope before sensitive file changes.

Question 381

A team upgrades its coding model and notices more flaky test fixes than before. Which approach should be recommended?

  1. Collect redacted command, diff, test and approval traces with correlation identifiers.
  2. Run a repository-specific regression benchmark before promoting the model in shared tooling.
  3. Enforce tests and status checks in the CI merge policy.
  4. Check in reviewed project-level Claude Code guidance and maintain it with code review.

Explanation

Correct Answer

B:

Quality gates on real codebases are more reliable than assuming a newly released model is an improvement. Verify measured coding-tool upgrades with actual repository access and CI controls; host-side controls are needed to guarantee measured coding-tool upgrades.

Incorrect Answers

A:

For a workflow whose primary weakness was auditable coding-agent activity, that proposal could be reasonable. Here the unmet obligation is measured coding-tool upgrades, and that requires the system to run a repository-specific regression benchmark before promoting the model in shared tooling.

C:

It is aimed at a different outcome, enforceable build acceptance rules. No evidence is given that it provides measured coding-tool upgrades; the needed mechanism is to run a repository-specific regression benchmark before promoting the model in shared tooling at the appropriate boundary.

D:

It emphasizes team-wide versioned coding instructions without closing the stated gap in measured coding-tool upgrades. By contrast, the remedy to run a repository-specific regression benchmark before promoting the model in shared tooling acts on the condition that makes this scenario fail.

Question 382

A developer uses a global project prompt containing database passwords to simplify automated tests. What is the best production design response?

  1. Collect redacted command, diff, test and approval traces with correlation identifiers.
  2. Remove secrets from prompt files and inject short-lived credentials through a vault-controlled runner.
  3. Enforce tests and status checks in the CI merge policy.
  4. Scope tool discovery and credentials by project role and environment.

Explanation

Correct Answer

B:

Instruction files are not secret stores; credential lifecycle belongs to trusted infrastructure. Verify secure credentials for AI-assisted development with actual repository access and CI controls; host-side controls are needed to guarantee secure credentials for ai-assisted development.

Incorrect Answers

A:

Consider its actual operating scope: auditable coding-agent activity. The requested assurance is secure credentials for ai-assisted development, which calls for an implementation that can remove secrets from prompt files and inject short-lived credentials through a vault-controlled runner, not merely for a related improvement.

C:

The proposed measure is aimed at enforceable build acceptance rules, not secure credentials for ai-assisted development. The architecture must instead remove secrets from prompt files and inject short-lived credentials through a vault-controlled runner; a separate change is still needed for secure credentials for ai-assisted development.

D:

The relevant trade-off is not primarily role- and environment-scoped tool access. A successful fix needs secure credentials for ai-assisted development; otherwise secure credentials for ai-assisted development remains unverified even after this change.

Question 383

Management measures coding-assistant value by counting its generated lines of code. Which change most directly satisfies this requirement?

  1. Run matched-task studies or controlled rollouts that account for project complexity.
  2. Measure merged changes, escaped defects and developer task time against a baseline.
  3. Evaluate findings by defect severity and actionable precision, not comment count.
  4. Require dependency analysis and limited diffs backed by regression coverage.

Explanation

Correct Answer

B:

Delivered maintainable changes are better productivity evidence than raw text volume. Measure engineering productivity through accepted outcomes in accepted changes and failures using comparable assignments, not in generated text or anecdotal speedups.

Incorrect Answers

A:

Its main effect concerns controlled evaluation of developer efficiency. The case explicitly hinges on engineering productivity through accepted outcomes; to resolve it reliably, the implementation has to measure merged changes, escaped defects and developer task time against a baseline, which this choice does not establish.

C:

The proposed action prioritizes high-value ai-assisted code review. That can coexist with the right solution, but it cannot replace engineering productivity through accepted outcomes: the application must still measure merged changes, escaped defects and developer task time against a baseline.

D:

The design concern is real but different: controlled scope for sensitive code edits. It does not impose the specific requirement for engineering productivity through accepted outcomes; a suitable solution must measure merged changes, escaped defects and developer task time against a baseline and demonstrate the result.

Question 384

A team asks Claude to generate API integration code but developers skip contract tests because output looks plausible. Which solution best resolves this operational gap?

  1. Apply software composition, license and vulnerability checks before merging.
  2. Require typed schema tests and integration checks before adoption.
  3. Use mutation testing or fault-injection checks to assess whether tests detect behavioral defects.
  4. Require dependency analysis and limited diffs backed by regression coverage.

Explanation

Correct Answer

B:

Generated code remains untrusted until it satisfies functional and security contracts. Measure tested AI-generated integration changes in accepted changes and failures using comparable assignments, not in generated text or anecdotal speedups.

Incorrect Answers

A:

It emphasizes governed dependency changes without closing the stated gap in tested ai-generated integration changes. By contrast, the remedy to require typed schema tests and integration checks before adoption acts on the condition that makes this scenario fail.

C:

Improving effective ai-generated test strength does not imply tested ai-generated integration changes. The selected design must require typed schema tests and integration checks before adoption, whereas this choice omits that direct protection or verification step.

D:

This recommendation targets controlled scope for sensitive code edits at a different point in the workflow. It leaves tested ai-generated integration changes outside the design contract and does not require the team to require typed schema tests and integration checks before adoption.

Question 385

A developer wants to refactor a complex service in one giant prompt that touches dozens of unrelated modules. What design decision would best fit these facts?

  1. Apply software composition, license and vulnerability checks before merging.
  2. Convert validated steps into a reusable reviewed Skill or automation template.
  3. Break the work into small reviewed diffs with tests after each change.
  4. Require dependency analysis and limited diffs backed by regression coverage.
  5. Require typed schema tests and integration checks before adoption.

Explanation

Correct Answer

C:

Incremental changes bound regression risk and provide clearer feedback to the coding model. Measure scoped AI-assisted refactoring in accepted changes and failures using comparable assignments, not in generated text or anecdotal speedups.

Incorrect Answers

A:

The relevant trade-off is not primarily governed dependency changes. A successful fix needs scoped ai-assisted refactoring; otherwise scoped ai-assisted refactoring remains unverified even after this change.

B:

The control would be justified if institutionalizing repeatable developer workflows were the limiting factor. In this scenario the decisive target is scoped ai-assisted refactoring; success requires an architecture that will break the work into small reviewed diffs with tests after each change.

D:

This shifts attention toward controlled scope for sensitive code edits when the relevant decision is about scoped ai-assisted refactoring. Unless the architecture can break the work into small reviewed diffs with tests after each change, the failure described in the stem remains unresolved.

E:

For a workflow whose primary weakness was tested ai-generated integration changes, that proposal could be reasonable. Here the unmet obligation is scoped ai-assisted refactoring, and that requires the system to break the work into small reviewed diffs with tests after each change.

Question 386

A new team repeats a successful deployment workflow manually from chat logs every sprint. What should the solution architect prioritize?

  1. Measure merged changes, escaped defects and developer task time against a baseline.
  2. Run matched-task studies or controlled rollouts that account for project complexity.
  3. Break the work into small reviewed diffs with tests after each change.
  4. Convert validated steps into a reusable reviewed Skill or automation template.

Explanation

Correct Answer

D:

A documented repeatable method turns individual prompting success into maintainable team practice. Measure institutionalizing repeatable developer workflows in accepted changes and failures using comparable assignments, not in generated text or anecdotal speedups.

Incorrect Answers

A:

The design concern is real but different: engineering productivity through accepted outcomes. It does not impose the specific requirement for institutionalizing repeatable developer workflows; a suitable solution must convert validated steps into a reusable reviewed Skill or automation template and demonstrate the result.

B:

This addresses controlled evaluation of developer efficiency; however, the problem asks for institutionalizing repeatable developer workflows. Without the safeguard to convert validated steps into a reusable reviewed Skill or automation template, the original failure mode is still possible.

C:

That choice would be responsive to scoped ai-assisted refactoring, but the reported defect is different. It leaves institutionalizing repeatable developer workflows unproved because there is no commitment to convert validated steps into a reusable reviewed Skill or automation template.

Question 387

A code review assistant adds many superficial comments but misses unsafe authentication logic. Which design choice best matches the stated constraint?

  1. Measure merged changes, escaped defects and developer task time against a baseline.
  2. Require dependency analysis and limited diffs backed by regression coverage.
  3. Evaluate findings by defect severity and actionable precision, not comment count.
  4. Require typed schema tests and integration checks before adoption.

Explanation

Correct Answer

C:

Review quality is determined by real detected problems and reviewer burden. Measure high-value AI-assisted code review in accepted changes and failures using comparable assignments, not in generated text or anecdotal speedups.

Incorrect Answers

A:

This recommendation targets engineering productivity through accepted outcomes at a different point in the workflow. It leaves high-value ai-assisted code review outside the design contract and does not require the team to evaluate findings by defect severity and actionable precision, not comment count.

B:

This is a valid architectural concern when optimizing controlled scope for sensitive code edits. Here it leaves high-value ai-assisted code review unaddressed, whereas the intended change must evaluate findings by defect severity and actionable precision, not comment count before the case can be considered resolved.

D:

By itself, this improves tested ai-generated integration changes, but the case is not asking the architect to optimize that dimension. The requirement is high-value ai-assisted code review, supported by an approach to evaluate findings by defect severity and actionable precision, not comment count.

Question 388

A company wants to quantify time savings from Claude Code but compares different teams with incompatible assignments. Which architecture adjustment is most appropriate?

  1. Convert validated steps into a reusable reviewed Skill or automation template.
  2. Require typed schema tests and integration checks before adoption.
  3. Run matched-task studies or controlled rollouts that account for project complexity.
  4. Measure merged changes, escaped defects and developer task time against a baseline.

Explanation

Correct Answer

C:

A credible productivity comparison separates tool effect from task and team differences. Measure controlled evaluation of developer efficiency in accepted changes and failures using comparable assignments, not in generated text or anecdotal speedups.

Incorrect Answers

A:

This shifts attention toward institutionalizing repeatable developer workflows when the relevant decision is about controlled evaluation of developer efficiency. Unless the architecture can run matched-task studies or controlled rollouts that account for project complexity, the failure described in the stem remains unresolved.

B:

This addresses tested ai-generated integration changes; however, the problem asks for controlled evaluation of developer efficiency. Without the safeguard to run matched-task studies or controlled rollouts that account for project complexity, the original failure mode is still possible.

D:

It is aimed at a different outcome, engineering productivity through accepted outcomes. No evidence is given that it provides controlled evaluation of developer efficiency; the needed mechanism is to run matched-task studies or controlled rollouts that account for project complexity at the appropriate boundary.

Question 389

AI-generated unit tests still pass after developers deliberately introduce behavioral faults. Separately, Claude-generated database migration code is being merged without any production-like rollback or data-integrity rehearsal. Which TWO engineering validation practices should be added? Choose TWO.

  1. Break the work into small reviewed diffs with tests after each change.
  2. Measure merged changes, escaped defects and developer task time against a baseline.
  3. Apply software composition, license and vulnerability checks before merging.
  4. Dry-run database migrations on representative data with integrity assertions and a tested rollback or recovery path.
  5. Use mutation testing or fault-injection checks to assess whether tests detect behavioral defects.

Explanation

Correct Answers

D:

Schema changes may pass ordinary code tests yet cause destructive transformations or nonreversible failures in real data. Practicing the migration and recovery against realistic fixtures establishes evidence beyond plausible generated code.

E:

Unit tests that still pass after realistic faults are introduced have weak defect-detection power. Mutation or fault-injection testing checks assertion strength rather than counting test cases; migration rehearsal addresses the distinct risk of generated changes to persistent data.

Incorrect Answers

A:

There is a legitimate use case for scoped ai-assisted refactoring, but it does not supply the test or enforcement needed for effective ai-generated test strength and production-like migration and recovery verification. The design needs to use mutation testing or fault-injection checks to assess whether tests detect behavioral defects.

B:

Consider its actual operating scope: engineering productivity through accepted outcomes. The requested assurance is effective ai-generated test strength and production-like migration and recovery verification, which calls for an implementation that can dry-run database migrations on representative data with integrity assertions and a tested rollback or recovery path, not merely for a related improvement.

C:

The alternative could reduce risk associated with governed dependency changes, yet the described behavior requires effective ai-generated test strength and production-like migration and recovery verification. That outcome depends on the capability to use mutation testing or fault-injection checks to assess whether tests detect behavioral defects.

Question 390

A coding model proposes a dependency upgrade with a new license and severe known vulnerability. Which approach should be recommended?

  1. Break the work into small reviewed diffs with tests after each change.
  2. Use mutation testing or fault-injection checks to assess whether tests detect behavioral defects.
  3. Require typed schema tests and integration checks before adoption.
  4. Apply software composition, license and vulnerability checks before merging.

Explanation

Correct Answer

D:

AI suggestions do not override the organization's supply-chain controls. Measure governed dependency changes in accepted changes and failures using comparable assignments, not in generated text or anecdotal speedups.

Incorrect Answers

A:

By itself, this improves scoped ai-assisted refactoring, but the case is not asking the architect to optimize that dimension. The requirement is governed dependency changes, supported by an approach to apply software composition, license and vulnerability checks before merging.

B:

Its main effect concerns effective ai-generated test strength. The case explicitly hinges on governed dependency changes; to resolve it reliably, the implementation has to apply software composition, license and vulnerability checks before merging, which this choice does not establish.

C:

The option might improve tested ai-generated integration changes; that remains insufficient for governed dependency changes. The case is resolved by governed dependency changes, which depends on a design that will apply software composition, license and vulnerability checks before merging.

Claude Certified Architect - Professional (CCAR-P) Practice Tests