Skill Scaffold

Determine output directory: Use skills/ relative to the current project root (where this skill is installed)

Create directory structure:

skills/<SKILL_NAME>/
├── SKILL.md
├── .claude-plugin/
│   └── plugin.json
└── README.md

If --with-python:

└── scripts/
    ├── <skill_name>.py    (snake_case version)
    └── requirements.txt

If --with-scripts:

└── scripts/
    └── run.sh

SKILL.md template — generate with:

---
name: <SKILL_NAME>
description: "<Ask user or use placeholder>"
argument-hint: '<SKILL_NAME> <args>'
allowed-tools: Bash, Read, Write, Edit
---

# <SKILL_NAME>

## Execution Logic

Check `$ARGUMENTS`:
- If **empty** → respond with usage and STOP.
- If **has arguments** → parse and execute.

## Task Execution

[TODO: Define skill logic here]

plugin.json template:

{
  "name": "<SKILL_NAME>",
  "description": "<same as SKILL.md description>",
  "version": "1.0.0",
  "author": { "name": "gp" },
  "repository": "https://github.com/gp/dev-environment-setup",
  "license": "MIT",
  "skills": ["./"]
}

Python stub (if --with-python):

#!/usr/bin/env python3
"""<SKILL_NAME> — [description]"""
import sys

def main():
    if len(sys.argv) < 2:
        print("Usage: <skill_name>.py <args>", file=sys.stderr)
        sys.exit(1)
    # TODO: implement
    print(f"Running {sys.argv[0]} with args: {sys.argv[1:]}")

if __name__ == "__main__":
    main()

README.md:

# <SKILL_NAME>

> [description]

## Usage

`/<skill-name> <args>`

## Installation

`npx skills i gp/dev-environment-setup --skill <skill-name>`

After creation, output:

✅ Skill "<SKILL_NAME>" created at skills/<SKILL_NAME>/

Files:
- SKILL.md (edit to define skill logic)
- .claude-plugin/plugin.json
- README.md
[- scripts/<file> (if --with-python or --with-scripts)]

Next: Edit SKILL.md to define your skill's behavior.