init

You are helping a user create a new Claude Code plugin. Follow this workflow step by step:

Step 1: Get GitHub Username and Fetch Profile

Use the AskUserQuestion tool to ask for:

• GitHub username - Will be used in CODEOWNERS and to fetch author information

Once you have the GitHub username, use WebFetch to fetch their profile:

• URL: https://api.github.com/users/{username}
• Extract the name field from the response
• If name is present and not null, use it as the author name
• If name is null or missing, use the GitHub username as the author name

Step 2: Collect Plugin Metadata

Ask the user for:

• Plugin name (kebab-case identifier, unique)
• Description (what does this plugin do?)
• Version (default: "1.0.0" if not specified)
• License (default: "MIT" if unsure)
• Homepage (optional - documentation URL)
• Repository (optional - source code URL, can default to GitHub repo if they want)
• Keywords (optional - array of tags for discoverability)
• Author email (optional - contact email)
• Author URL (optional - personal website/profile)

Step 3: Create Plugin Directory

Create the directory structure at the top level:

./{plugin-name}/
./{plugin-name}/.claude-plugin/ # required
./{plugin-name}/commands/ # optional - only if 1 or more commands added
./{plugin-name}/agents/ # optional - only if 1 or more agents added
./{plugin-name}/hooks/ # optional - only if 1 or more hooks added
./{plugin-name}/skills/ # optional - only if 1 or more skills added
./{plugin-name}/mcp-servers/ # optional - only if 1 or more MCP servers added

Step 4: Select Components to Create

Use AskUserQuestion with multiSelect=true to ask what components they want to create:

• Slash Command
• Agent (Subagent)
• Hook
• Skill
• MCP Server

Step 5: For Each Component Type, Collect Details

If they selected "Slash Command":

Ask these questions (you can ask multiple in one AskUserQuestion call):

1. Command name (kebab-case, will be invoked as /plugin-name:command-name)
2. Command description (one-line summary)
3. What should this command do? (detailed explanation of the command's purpose and behavior)
4. What files/resources will it need to read or modify?
5. Should it use any specific tools? (e.g., Bash, Read, Edit, Grep, etc.)