Custom slash commands and skills
How .claude/commands/ files and SKILL.md skills create the same /name commands, where project and personal skills live and which wins a name clash, and what argument-hint, context: fork, allowed-tools, disallowed-tools and disable-model-invocation actually do. Also when a workflow belongs in a skill rather than CLAUDE.md.
Create and configure custom slash commands and skills
~8 min read · 10 practice questions ·
What the exam expects
- Create a team-wide slash command by committing a
.claude/commands/<name>.mdfile or a.claude/skills/<name>/SKILL.mdskill, and a personal variant under~/.claude/skills/with a different name. - Configure SKILL.md frontmatter:
argument-hintfor expected parameters,context: forkfor isolation,allowed-toolsanddisallowed-toolsfor tool access, anddisable-model-invocationfor side-effect workflows. - Use
context: forkfor verbose or exploratory task skills, and recognize when it breaks a reference-only skill. - Choose skills (on-demand, task-specific) over CLAUDE.md (always loaded, universal standards).
- Make an Agent SDK agent discover project skills.
Key terms: skill (a directory with a SKILL.md file that Claude or a person can invoke as /name); frontmatter (YAML between --- markers that configures the skill); forked skill (a skill run as an isolated subagent via context: fork); progressive disclosure (description always in context, body on invoke, supporting files when needed); pre-approval (tools usable without a permission prompt).
Key facts
- Detail
.claude/commands/deploy.mdand.claude/skills/deploy/SKILL.mdboth create/deploy
- Detail
- No forced migration; same frontmatter except
nameandpaths
- Detail
- Enterprise > personal > project; skill beats a command file
- Detail
- Name and description, about 100 tokens per skill
- Detail
- Enters as one message, stays across turns, not re-read
allowed-tools- Detail
- Pre-approves for the invoking turn; does not restrict
disallowed-tools- Detail
- Removes tools while the skill is active
context: fork input- Detail
- Skill content as prompt; no conversation history
argument-hint- Detail
- Autocomplete hint only
- Detail
- Filesystem only, via
settingSources; no registration API
- Why does a personal
~/.claude/skills/deploy/SKILL.mdhide the team's/deployfrom its author? - When should a skill use
context: fork? - Why can many skills be installed without a large context penalty?
- Which frontmatter field removes Bash while a skill is active?
- What must an SDK agent's
setting_sourcesinclude to discover.claude/skills/?
The questions are free to read. Answers with their sources and self-grading are included with the pack.
Keep reading with the Claude Certified Architect - Foundations pack
- Core concepts
- Exam traps
- Worked scenario
The pack includes:
- All 30 lessons of the Claude Certified Architect - Foundations guide, with reading progress
- Interactive decision walks and step-by-step sequence steppers
- Inline practice questions and one-click topic practice (10 questions for this lesson)
- An official source cited for every claim
- An AI tutor that answers from the guide
- Unlimited timed practice exams
Practice for the Claude Certified Architect - Foundations exam
Certifyforce is an independent study resource and is not affiliated with, endorsed by or sponsored by Anthropic. Claude is a trademark of Anthropic.