CCARF Practice Question

Paper 1 · Question 54

D3 Claude Code Configuration & Workflows · Paper 1

A documentation team creates a /generate-api-docs skill that reads source code files and produces Markdown API reference pages. The skill generates verbose output (200+ lines per endpoint) and should be available to every team member who clones the repository. How should the skill be configured?

• A.Create a SKILL.md in ~/.claude/skills/ with context: fork frontmatter, and instruct each team member to copy it locally
• B.Create a SKILL.md in .claude/skills/ with context: fork, so it is shared via git and isolates the verbose output
• C.Add the documentation generation instructions to the root CLAUDE.md so they load automatically in every session
• D.Create a SKILL.md in .claude/skills/ without any frontmatter, relying on the team to manually manage context overflow

Show answer & explanations

Correct answer: B

A: ~/.claude/skills/ is user-scoped and not version-controlled. Requiring manual copying defeats the purpose of team-shared configuration and creates maintenance drift.

B: .claude/skills/ is project-scoped and version-controlled, making it available to every developer who clones the repository. context: fork isolates the verbose 200+ line output per endpoint into a forked context, preventing it from consuming the main conversation's context budget.

C: CLAUDE.md is for always-loaded standards, not for on-demand command definitions. Loading verbose generation instructions in every session wastes context tokens when the team is doing non-documentation work.

D: Without context: fork, the verbose output (200+ lines per endpoint across many endpoints) would flood the main conversation context. The context: fork frontmatter is essential for isolating bulky output.

Related Claude Code Configuration & Workflows questions

More: Paper 1 · All practice questions