slash-command-creator

# Slash Command Creator

Create effective Claude Code slash commands - reusable prompt templates stored as Markdown files.

## Quick Reference

| Scope | Location | Label in /help |
|-------|----------|----------------|
| Project | `.claude/commands/*.md` | (project) |
| Personal | `~/.claude/commands/*.md` | (user) |

## File Structure

```markdown
---
description: Brief description shown in /help
argument-hint: <required> [optional]
allowed-tools: Bash(git:*), Read, Edit
model: claude-sonnet-4-20250514
disable-model-invocation: false
---

Your prompt template here with $ARGUMENTS or $1, $2, etc.
```

## Variables

| Variable | Description | Example |
|----------|-------------|---------|
| `$ARGUMENTS` | All args as single string | `/cmd foo bar` -> `"foo bar"` |
| `$1`, `$2`... | Positional args | `/cmd foo bar` -> `$1="foo"`, `$2="bar"` |
| `@path/file` | Include file contents | `Review @src/main.py` |
| `` `!command` `` | Execute bash, include output | `` `!git status` `` |

## Frontmatter Fields

| Field | Purpose | Default |
|-------|---------|---------|
| `description` | Shown in /help, enables discoverability | First line of prompt |
| `argument-hint` | Shows in autocomplete | None |
| `allowed-tools` | Restrict tool access | Inherits from conversation |
| `model` | Override model | Inherits from conversation |
| `disable-model-invocation` | Prevent auto-invocation | `false` |

## Creation Process

1. **Identify the workflow** - What repetitive task needs automation?
2. **Choose scope** - Project-specific or personal?
3. **Design the prompt** - Keep it focused on one task
4. **Add variables** - Use `$ARGUMENTS` for flexibility
5. **Set frontmatter** - Add description and hints
6. **Test and iterate** - Refine based on usage

## When to Use Slash Commands vs Skills

**Use slash commands for:**
- Quick, single-file prompts
- Frequent manual invocations
- Simple templates and reminders
- Team workflows in version control

**Use Skills instead for:**
- Multi-fil