CCDV-F Practice Questions: Claude Code Domain

Correct answer (A): Repository initialization helps Claude Code operate with project context instead of treating each interaction as isolated. Adding durable project guidance, typically through repository files such as CLAUDE.md, gives Claude Code shared conventions to use in future sessions. The problem described is missing project setup, not model capability.

Why the other options are wrong:
- Option B: Changing models does not address the missing repository context and project guidance described in the session.
- Option C: A one-time prompt may help one response, but it does not create durable project-aware behavior for future work.
- Option D: Personal memory is not the best mechanism for shared, versioned repository conventions.

Correct answer (A): settings.json is the appropriate place for durable Claude Code configuration behavior. The scenario is about operational settings that should apply consistently without a repeated setup prompt, so a repository or project configuration file is the best fit. CLAUDE.md is better for natural-language project guidance rather than configurable Claude Code behavior.

Why the other options are wrong:
- Option B: CLAUDE.md is useful for repository instructions, but the scenario asks for durable operational configuration rather than coding guidance.
- Option C: Agent Memory is not the best fit for team-wide project configuration because it is not a versioned repository configuration mechanism.
- Option D: Manual setup remains error-prone and does not satisfy the requirement for consistent behavior without repeated prompts.

Correct answer (A): Claude Code supports a hierarchy of CLAUDE.md files, and file placement matters. A root CLAUDE.md can provide shared repository guidance, while a nested CLAUDE.md can add package- or directory-specific instructions. In the payments directory, the best interpretation is to apply the shared root guidance together with the payments-specific guidance.

Why the other options are wrong:
- Option B: Root guidance is relevant, but hierarchy allows nested guidance to add more specific instructions for a subdirectory.
- Option C: The closer file matters, but the shared root guidance is still intended to apply across the repository.
- Option D: CLAUDE.md is designed to provide durable project guidance without requiring the developer to paste it manually into each session.

Correct answer (A): settings.json is the place for Claude Code configuration, while CLAUDE.md is better suited for human-readable project guidance and development conventions. The instruction describes how to work on billing code, not a Claude Code configuration setting.

Why the other options are wrong:
- Option B: Renaming the key does not change the fact that settings.json is the wrong place for project coding guidance.
- Option C: A custom slash command is useful for repeatable invocations, but this is persistent guidance that should apply whenever relevant code is edited.
- Option D: Session history is not durable, reviewable team guidance and will not reliably help other contributors.

Correct answer (A): A maintainable Claude Code repository setup separates concerns: CLAUDE.md for durable project guidance, custom commands for repeatable project-specific workflows, and settings.json for configuration. This supports consistency across contributors and keeps behavior easier to review as the repository evolves.

Why the other options are wrong:
- Option B: settings.json is not the right place for all project conventions and workflows; it should not replace CLAUDE.md guidance or custom commands.
- Option C: Per-engineer Agent Memory would make shared repository behavior inconsistent and less reviewable across the team.
- Option D: Manual prompts are not durable or repeatable and would undermine consistency across contributors.

Correct answer (A): CLAUDE.md is the appropriate Claude Code mechanism for durable, human-readable project or repository instructions. It avoids repeated ad hoc prompting and makes conventions easier for the team to review and maintain across sessions and contributors.

Why the other options are wrong:
- Option B: Agent Memory can retain context or preferences, but shared project conventions should be explicit and reviewable in repository guidance rather than hidden per developer.
- Option C: settings.json is for Claude Code configuration, not for storing human-readable coding standards as comments or policy guidance.
- Option D: Repeating the guidance manually is exactly the brittle workflow the team is trying to eliminate.

Correct answer (A): Streaming mode is the right fit when operators or automation need incremental visibility into Claude Code output rather than waiting only for a final response. The requirement is specifically about observing progress during a long non-interactive run.

Why the other options are wrong:
- Option B: This directly conflicts with the requirement to see progress during the run.
- Option C: Agent Memory cannot provide incremental runtime visibility into a non-interactive execution.
- Option D: A custom slash command may make invocation repeatable, but it does not address the need to stream progress output.

Correct answer (A): Headless mode is appropriate for non-interactive or automation-oriented Claude Code execution where no developer needs an interactive chat session. The requirement explicitly calls for unattended operation that produces an artifact.

Why the other options are wrong:
- Option B: An interactive session conflicts with the requirement that no developer be present and no chat prompts be required.
- Option C: Agent Memory does not define the operating mode for unattended automation and would not by itself run the job.
- Option D: A nested CLAUDE.md may provide guidance, but it does not replace the need for a non-interactive execution pattern.

Correct answer (A): Headless mode is appropriate for non-interactive or scripted Claude Code execution. The workload is scheduled, unattended, and expected to produce an artifact and exit, so headless execution is a better operational fit than an interactive session that depends on a human.

Why the other options are wrong:
- Option B: Interactive sessions are useful when a developer is guiding work step by step, but this job is unattended automation.
- Option C: Agent Memory can preserve context, but it does not provide the non-interactive execution mode required by the scheduler.
- Option D: A custom slash command may standardize a task, but the key requirement is unattended scripted execution.

Correct answer (A): CLAUDE.md is the appropriate durable, repository-scoped mechanism for project guidance such as coding conventions, preferred workflows, and expectations that should be shared by the team. Because the requirement is reviewable, source-controlled, and applied across developers, repository guidance belongs in CLAUDE.md rather than in transient prompts or personal memory.

Why the other options are wrong:
- Option B: Agent Memory may help with personal or persistent assistant context, but it is not the best substitute for versioned, team-wide repository instructions.
- Option C: Repeating instructions in each session is workable temporarily, but it is not durable, reviewable, or consistent across the team.
- Option D: settings.json is for Claude Code configuration behavior, not the best place for natural-language repository coding standards.