create-skill

Create a New Skill

Create a new skill called $0 in the $1 plugin.

Instructions

1. Create the directory: plugins/$1/skills/$0/
2. Create plugins/$1/skills/$0/SKILL.md using the template below
3. If a description was provided, use it: $2

SKILL.md Template

Use this exact structure for the new skill file:

---
name: <skill-name>
description: "<One line: what it does and when Claude should use it. Max 250 chars.>"
user-invocable: true
argument-hint: "<placeholder args the user passes, e.g. [target] [options]>"
---

# <Skill Title>

<Clear, imperative instructions for Claude. Write as if briefing a senior engineer.>

## When to use

<1-3 bullet points describing trigger conditions>

## Inputs

- `$ARGUMENTS` — full argument string from the user
- `$0` — first argument (usually the target)
- `$1` — second argument (usually an option or modifier)

## Steps

1. <Step one>
2. <Step two>
3. <Step three>

## Output format

<What the skill should produce: code, markdown doc, structured analysis, etc.>

## Constraints

- <Guard rails, things to avoid, scope limits>

Frontmatter Reference

Field	Required	Description
name	Yes	Kebab-case identifier
description	Yes	When Claude should invoke this (max 250 chars)
user-invocable	No	true (default) = user can call via /skill-name
disable-model-invocation	No	true = only user can trigger, Claude won't auto-invoke
argument-hint	No	Shown in autocomplete, e.g. [file] [--verbose]
allowed-tools	No	Space-separated tool names Claude can use without prompts
model	No	Force a specific model: sonnet, opus, haiku
effort	No	low, medium, high, max
context	No	fork = run in isolated subagent
paths	No	Glob patterns for auto-loading

Tips

• Keep descriptions under 250 characters — they're used for skill matching
• Use $ARGUMENTS for dynamic input; $0, $1 for positional args
• Use disable-model-invocation: true for destructive or opinionated skills
• Add a ## Constraints section to prevent scope creep
• Reference files with ${CLAUDE_SKILL_DIR} for co-located templates