Back to Skills

Plugin Creator

jeremylongshore
Updated Today
25 views
927
113
927
View on GitHub
Metaaiautomation

About

This Claude Skill automatically scaffolds new Claude Code plugins with proper structure and marketplace integration when triggered by phrases like "create plugin" or "plugin from template." It handles the complete setup process including directory creation, file generation, and catalog updates specifically for the claude-code-plugins repository workflow. Developers should use it to quickly generate validated plugin templates with all required components.

Documentation

Plugin Creator

Purpose

Automatically scaffolds new Claude Code plugins with complete directory structure, required files, proper formatting, and marketplace catalog integration - specifically optimized for the claude-code-plugins repository.

Trigger Keywords

  • "create plugin" or "new plugin"
  • "plugin from template"
  • "scaffold plugin"
  • "generate plugin"
  • "add new plugin to marketplace"

Plugin Creation Process

When activated, I will:

  1. Gather Requirements

    • Plugin name (kebab-case)
    • Category (productivity, security, devops, etc.)
    • Type (commands, agents, skills, MCP, or combination)
    • Description and keywords
    • Author information
  2. Create Directory Structure

    plugins/[category]/[plugin-name]/
    ├── .claude-plugin/
    │   └── plugin.json
    ├── README.md
    ├── LICENSE
    └── [commands|agents|skills|hooks|mcp]/
    
  3. Generate Required Files

    • plugin.json with proper schema (name, version, description, author)
    • README.md with comprehensive documentation
    • LICENSE (MIT by default)
    • Component files based on type
  4. Add to Marketplace Catalog

    • Update .claude-plugin/marketplace.extended.json
    • Run npm run sync-marketplace automatically
    • Validate catalog schema
  5. Validate Everything

    • Run ./scripts/validate-all.sh on new plugin
    • Check JSON syntax with jq
    • Verify frontmatter in markdown files
    • Ensure scripts are executable

Plugin Types Supported

Commands Plugin

  • Creates commands/ directory
  • Generates example command with proper frontmatter
  • Includes /demo-command example

Agents Plugin

  • Creates agents/ directory
  • Generates example agent with capabilities
  • Includes model specification

Skills Plugin

  • Creates skills/skill-name/ directory
  • Generates SKILL.md with proper format
  • Includes trigger keywords and allowed-tools

MCP Plugin

  • Creates src/, dist/, mcp/ directories
  • Generates TypeScript boilerplate
  • Includes package.json with MCP SDK
  • Adds to pnpm workspace

Full Plugin

  • Combines all types
  • Creates complete example structure
  • Ready for customization

File Templates

plugin.json Template

{
  "name": "plugin-name",
  "version": "1.0.0",
  "description": "Clear description",
  "author": {
    "name": "Author Name",
    "email": "[email protected]"
  },
  "repository": "https://github.com/jeremylongshore/claude-code-plugins",
  "license": "MIT",
  "keywords": ["keyword1", "keyword2"]
}

Command Template

---
name: command-name
description: What this command does
model: sonnet
---

# Command Title

Instructions for Claude...

Skill Template

---
name: Skill Name
description: What it does AND when to use it
allowed-tools: Read, Write, Grep
---

# Skill Name

## Purpose
[What this skill does]

## Trigger Keywords
- keyword1
- keyword2

## Instructions
[Step-by-step for Claude]

Marketplace Integration

I automatically:

  1. Add plugin entry to marketplace.extended.json
  2. Run npm run sync-marketplace to update CLI catalog
  3. Validate both catalogs with jq
  4. Check for duplicate names
  5. Verify source paths exist

Validation Steps

After creation:

  • ✅ All required files present
  • ✅ Valid JSON (plugin.json, catalogs)
  • ✅ Proper frontmatter in markdown
  • ✅ Scripts executable (chmod +x)
  • ✅ No duplicate plugin names
  • ✅ Category is valid
  • ✅ Keywords present

Repository-Specific Features

For claude-code-plugins repo:

  • Follows exact directory structure
  • Uses correct marketplace slug (claude-code-plugins-plus)
  • Includes proper LICENSE file
  • Adds to correct category folder
  • Validates against existing plugins
  • Updates version in marketplace

Output

I provide:

✅ Created plugin: plugin-name
📁 Location: plugins/category/plugin-name/
📝 Files created: 8
🔍 Validation: PASSED
📦 Marketplace: UPDATED
✨ Ready to commit!

Next steps:
1. Review files in plugins/category/plugin-name/
2. Customize README.md and component files
3. Run: git add plugins/category/plugin-name/
4. Run: git commit -m "feat: Add plugin-name plugin"

Examples

User says: "Create a new security plugin called 'owasp-scanner' with commands"

I automatically:

  1. Create directory: plugins/security/owasp-scanner/
  2. Generate plugin.json, README, LICENSE
  3. Create commands/ with example
  4. Add to marketplace.extended.json
  5. Sync marketplace.json
  6. Validate all files
  7. Report success

User says: "Scaffold a Skills plugin for code review"

I automatically:

  1. Create directory with skills/ subdirectories
  2. Generate SKILL.md templates
  3. Add trigger keywords for code review
  4. Add to marketplace
  5. Validate and report

Quick Install

/plugin add https://github.com/jeremylongshore/claude-code-plugins-plus-skills/tree/main/plugin-creator

Copy and paste this command in Claude Code to install this skill

GitHub 仓库

jeremylongshore/claude-code-plugins-plus-skills
Path: backups/plugin-enhancements/plugin-backups/skills-powerkit_20251020_012455/skills/plugin-creator
aiautomationclaude-codedevopsmarketplacemcp

Related Skills

sglang

Meta

SGLang is a high-performance LLM serving framework that specializes in fast, structured generation for JSON, regex, and agentic workflows using its RadixAttention prefix caching. It delivers significantly faster inference, especially for tasks with repeated prefixes, making it ideal for complex, structured outputs and multi-turn conversations. Choose SGLang over alternatives like vLLM when you need constrained decoding or are building applications with extensive prefix sharing.

View skill

evaluating-llms-harness

Testing

This Claude Skill runs the lm-evaluation-harness to benchmark LLMs across 60+ standardized academic tasks like MMLU and GSM8K. It's designed for developers to compare model quality, track training progress, or report academic results. The tool supports various backends including HuggingFace and vLLM models.

View skill

llamaguard

Other

LlamaGuard is Meta's 7-8B parameter model for moderating LLM inputs and outputs across six safety categories like violence and hate speech. It offers 94-95% accuracy and can be deployed using vLLM, Hugging Face, or Amazon SageMaker. Use this skill to easily integrate content filtering and safety guardrails into your AI applications.

View skill

langchain

Meta

LangChain is a framework for building LLM applications using agents, chains, and RAG pipelines. It supports multiple LLM providers, offers 500+ integrations, and includes features like tool calling and memory management. Use it for rapid prototyping and deploying production systems like chatbots, autonomous agents, and question-answering services.

View skill