Claude Code Slash Commands

Reusable, parameterized prompts you invoke with /command. Cheap, fast way to encode repeatable workflows.

Where they live

~/.claude/commands/ — user-global commands.
<repo>/.claude/commands/ — project-shared (committed to repo).
Each file is a markdown prompt. Filename = command name (e.g., review.md → /review).

Anatomy

<repo>/.claude/commands/review.md:

---
description: Review the current diff for correctness and security.
allowed-tools: Bash(git diff:*), Bash(git status), Read, Grep
---

Review the staged and unstaged diff in this repo.

Focus on:
- correctness bugs
- security issues (injection, auth, secrets)
- missing tests

Output a markdown checklist of findings, severity-tagged.

Frontmatter is optional. Useful fields:

description — shown in / autocomplete.
allowed-tools — narrow tool set for this command.
argument-hint — placeholder shown to user.

Arguments

In the markdown body, $ARGUMENTS is replaced with whatever the user typed after the command name.

Generate a unit test for the function $ARGUMENTS.

Usage: /test-for src/utils/format.ts:formatDate

Bash interpolation

Use backticks to embed shell output:

The current branch is `git branch --show-current`. Review since `git merge-base HEAD main`.

Slash command vs Skill vs Subagent

Need	Use
Reusable parameterized prompt	Slash command
Larger reusable capability with files/binaries	Skill
Isolated context for a delegated task	Subagent

CCA-F angle

Recognize when to use a slash command vs a skill vs a subagent. Slash commands are stateless prompt templates; skills include resources and instructions; subagents are isolated execution.