command-creator

Original：🇺🇸 English

Translated

Create Claude Code custom slash commands with proper structure, frontmatter, and best practices. Use this skill whenever the user wants to create a new command, add a slash command, build a custom command, or mentions "create-command", "new command", "add command", or "make a command" for Claude Code. Also trigger when the user wants to turn a workflow into a reusable command.

2installs

Sourcefeiskyer/claude-code-settings

Added on2026-05-19

NPX Install

npx skill4agent add feiskyer/claude-code-settings command-creator

SKILL.md Content

View Translation Comparison →

Command Creator

Create Claude Code custom slash commands with proper structure and best practices.

Understanding Commands

Claude Code commands are Markdown files with YAML frontmatter that define reusable workflows invoked via

/command-name

. They live in specific locations depending on scope:

Project-level:
```
.claude/commands/
```
in the project root (shared via git)
User-level:
```
~/.claude/commands/
```
(personal, available everywhere)
Categorized: Nest under subdirectories for namespacing (e.g.,
```
gh/review-pr.md
```
becomes
```
/gh:review-pr
```
)

Process

1. Clarify Intent

Before writing anything, understand:

What should the command do? (single clear purpose)
Who uses it — just this user, or the whole team? (determines project vs user-level)
Does it need arguments? What kind?
What tools does it need access to?
Is there a category it belongs to? (e.g.,
```
gh
```
for GitHub,
```
cc
```
for Claude Code meta-commands)

2. Write the Command

Every command file follows this structure:

markdown

---
description: Brief description shown in command list
argument-hint: [expected-arguments]
allowed-tools: Tool1, Tool2, Bash(prefix:*)
---

# Command Name

What this command does and when to use it.

## Process:

Step-by-step instructions for the agent to follow.

## Your Task:

Act on "$ARGUMENTS" following these guidelines.

Frontmatter fields

Field	Required	Purpose
`description`	Yes	Short description shown when listing commands
`argument-hint`	No	Hint for expected arguments (shown in autocomplete)
`allowed-tools`	No	Restrict which tools the command can use

Key conventions

$ARGUMENTS
is replaced with whatever the user types after the command name. Always reference it in the "Your Task" section so the command acts on user input.
allowed-tools
uses patterns: exact names (
```
Read
```
,
```
Write
```
), or prefix globs for Bash (
```
Bash(git:*)
```
,
```
Bash(npm:*)
```
).
Keep commands focused and single-purpose — one command, one job.
Write instructions in the imperative form ("Analyze the code", not "You should analyze the code").
Include concrete examples of usage and expected behavior.

3. Choose the Right Location

Scope	Path	When to use
Project (shared)	`.claude/commands/`	Team workflows, project-specific tasks
Project (categorized)	`.claude/commands/<category>/`	Grouped commands (e.g., `gh/` , `db/` )
User (personal)	`~/.claude/commands/`	Personal productivity, cross-project tools

4. Validate

After creating the command file:

Verify the YAML frontmatter parses correctly (no syntax errors)
Confirm the file is in the right directory
Check that
```
$ARGUMENTS
```
is referenced if the command accepts input
Ensure
```
allowed-tools
```
includes everything the command needs

Example: A Simple Command