FEFreeExamDumps.in

CCA-F Practice Questions — Page 10

Question 91

Open question ↗

Scenario: Claude Code Configuration and Workflows -

You want a `/review` slash command that is automatically available to the whole team when they clone the repository.

Where should the command file live?

  • A.In the project's `.claude/commands/` (or `.claude/skills/`), which is committed to version control.
  • B.In `~/.claude/commands/` on each developer's machine.
  • C.In the root CLAUDE.md as a prose instruction.
  • D.In a `.claude/config.json` commands array.

Question 92

Open question ↗

Scenario: Claude Code Configuration and Workflows -

A code-analysis skill produces very verbose output that pollutes the main session, and it should only be allowed to read files (never write or delete).

Which SKILL.md frontmatter settings address both needs?

  • A.`context: fork` to isolate it in a subagent, and `allowed-tools: ["Read", "Grep", "Glob"]` to restrict capabilities.
  • B.`context: main` and `allowed-tools: ["*"]`.
  • C.`argument-hint` set to a long description of the analysis.
  • D.No frontmatter; rely on the prompt to tell the skill not to write files.

Question 93

Open question ↗

Scenario: Claude Code Configuration and Workflows -

You must migrate a library used across 45+ files and choose between several plausible refactoring approaches. The codebase is unfamiliar.

Which mode should you start in?

  • A.Planning mode: explore with Read/Grep/Glob, understand dependencies, and produce a plan for approval before changing anything.
  • B.Direct execution: start editing files immediately and adjust as problems appear.
  • C.Direct execution with one long up-front instruction covering all 45 files.
  • D.Skip planning and run a single batch refactor command.

Question 94

Open question ↗

Scenario: Claude Code Configuration and Workflows -

A clear stack trace points to a single null-check bug in one file, and the fix is obvious and unambiguous.

What is the appropriate workflow?

  • A.Direct execution—the change is small, well-understood, and confined to one file.
  • B.Planning mode with a full codebase exploration first.
  • C.Fork the session to compare two ways of adding the null check.
  • D.Spawn an Explore subagent to map the whole module before fixing.

Question 95

Open question ↗

Scenario: Claude Code Configuration and Workflows -

A CI pipeline runs `claude "Review this PR for security issues"` and the job hangs indefinitely without producing output.

What is the correct invocation for CI?

  • A.`claude -p "Review this PR for security issues"` to run non-interactively, print to stdout, and exit.
  • B.Set an environment variable `CLAUDE_HEADLESS=true`.
  • C.Add `--batch` to the command.
  • D.Pipe `yes |` into the command to auto-confirm prompts.

Question 96

Open question ↗

Scenario: Claude Code Configuration and Workflows -

In CI you need Claude Code's review output to be machine-parseable so a script can post inline PR comments and validate the structure.

Which flags should you use?

  • A.`--output-format json` together with `--json-schema` to validate the output structure.
  • B.`--verbose` so the script can scrape the logs.
  • C.`--markdown` so comments render nicely.
  • D.No flags; parse the default prose output with regex.

Question 97

Open question ↗

Scenario: Claude Code Configuration and Workflows -

In a CI workflow, the same Claude session that generated a feature is then asked to review its own code. Reviewers notice it rarely challenges its own decisions and misses issues.

What is the recommended practice?

  • A.Use an independent Claude instance (without the generation context) to perform the review.
  • B.Keep the same session but raise the temperature for the review step.
  • C.Ask the same session to review twice and merge the results.
  • D.Disable CLAUDE.md during the review so the session forgets its reasoning.

Question 98

Open question ↗

Scenario: Claude Code Configuration and Workflows -

Claude Code re-reviews a PR after new commits and re-posts all the same comments it already made on the previous run, frustrating developers with duplicate feedback.

How do you prevent duplicate comments?

  • A.Include prior review results in context and instruct Claude to report only new or unresolved issues.
  • B.Delete all previous comments before each run.
  • C.Only review the PR once and never re-review after new commits.
  • D.Lower `max_tokens` so fewer comments fit.

Question 99

Open question ↗

Scenario: Claude Code Configuration and Workflows -

During a long investigation the context window is filling with verbose tool output, and you want to compress prior history to keep working—while being aware of a known risk.

Which statement about `/compact` is correct?

  • A.`/compact` summarizes prior history to free context, but exact numbers, dates, and specific details can be lost in summarization.
  • B.`/compact` losslessly stores the full history to disk and reloads it on demand.
  • C.`/compact` permanently deletes the conversation and starts a brand-new session.
  • D.`/compact` increases the model's context window size for the session.

Question 100

Open question ↗

Scenario: Prompt Engineering and Structured Output -

A code-comment checker uses the instruction "Check comments for accuracy and be conservative." It produces many low-value flags (stylistic nitpicks, minor wording) and developers stop trusting it.

How should the prompt be rewritten?

  • A.Give explicit criteria: flag a comment ONLY if it contradicts code behavior, references a non-existent symbol, or a TODO refers to an already-fixed bug—and list what NOT to flag.
  • B.Add "be even more conservative and precise" to the instruction.
  • C.Raise the model temperature to encourage variety in findings.
  • D.Ask the model to flag everything and let developers filter.