claude-code-plugin-automations

内容来源：README.md（说明文档） · 原始地址 · 查看安装指南

原始内容

Claude Code Plugins

Claude Marketplace Plugins License Validate Plugins Conventional Commits

A collection of plugins and meta-agents for Claude Code, Anthropic's official CLI for Claude. These plugins extend Claude's capabilities with specialized agents, skills, commands, and hooks.

CI/CD Pipeline: Automated validation, changelog management, and semantic versioning. See CI/CD Guide →

Now available on Claude Marketplace! Install with one command or browse plugins directly from Claude Code.

🚀 What is Claude Code?

Claude Code is an interactive CLI tool that helps with software engineering tasks. It supports:

  • Agents: Specialized subagents for delegated tasks
  • Skills: Auto-invoked expertise modules
  • Slash Commands: User-triggered workflows
  • Hooks: Event-driven automation
  • Plugins: Bundled collections of the above

📦 Available Plugins

Agent Builder (Meta-Agent)

A meta-agent for building other Claude agents!

The Agent Builder plugin provides comprehensive tools for creating Claude Code extensions. It includes expert agents, auto-invoked skills, validation scripts, and templates for all component types.

Features:

  • 🤖 Meta-architect agent for expert guidance
  • 🛠️ 4 specialized skills that auto-invoke when building components
  • ⚡ 5 slash commands for quick scaffolding
  • ✅ Validation scripts for schema compliance
  • 📝 Complete template library
  • 🔒 Security-focused best practices

Quick Start:

# Create a new agent
/claude-component-builder:agents:new code-reviewer

# Create a new skill
/claude-component-builder:skills:new analyzing-data

# Create a new command
/claude-component-builder:commands:new run-tests

# Create a new hook
/claude-component-builder:hooks:new validate-writes

# Create a complete plugin
/claude-component-builder:plugins:new my-tools

See full documentation →

Self-Improvement Plugin

Claude critiquing Claude - A feedback loop for continuous improvement!

The Self-Improvement plugin enables Claude to critique its own work, identify quality issues, and create feedback loops for continuous learning. It includes a self-critic agent and skills for quality analysis, improvement suggestions, and feedback loops.

Features:

  • 🔍 Self-critic agent for honest, constructive feedback
  • 📊 Quality analysis across 6 dimensions (correctness, completeness, clarity, efficiency, security, usability)
  • 💡 Actionable improvement suggestions with before/after examples
  • 🔄 Feedback loop systems for continuous learning
  • ⚡ Quick quality checks and comprehensive reviews
  • 📈 Pattern tracking and learning from mistakes

Quick Start:

# Comprehensive review of recent work
/review-my-work

# Quick quality assessment
/quality-check

Use Cases:

  • Catch bugs before delivery
  • Learn from recurring mistakes
  • Iterative refinement of solutions
  • Security hardening through self-review
  • Communication improvement

See full documentation →

📥 Installation

From Claude Marketplace (Recommended)

Install directly from the Claude Code marketplace:

# Add the marketplace to your Claude Code settings
claude marketplace add https://github.com/C0ntr0lledCha0s/claude-code-plugin-automations

# Install specific plugins
claude plugin install claude-component-builder
claude plugin install self-improvement

Or add to your .claude/settings.json:

{
  "marketplaces": [
    "https://github.com/C0ntr0lledCha0s/claude-code-plugin-automations"
  ],
  "plugins": [
    "claude-component-builder",
    "self-improvement"
  ]
}

For All Projects (User-Level)

Install plugins to your user-level Claude directory:

# Clone the repository
git clone https://github.com/C0ntr0lledCha0s/claude-code-plugin-automations.git
cd claude-code-plugin-automations

# Link plugins to Claude user directory
mkdir -p ~/.claude/plugins
ln -s $(pwd)/claude-component-builder ~/.claude/plugins/claude-component-builder
ln -s $(pwd)/self-improvement-plugin ~/.claude/plugins/self-improvement

For a Specific Project

Install plugins at the project level:

# Navigate to your project
cd /path/to/your/project

# Clone the plugins
git clone https://github.com/C0ntr0lledCha0s/claude-code-plugin-automations.git

# Link components to project .claude directory
mkdir -p .claude/{agents,skills,commands}

ln -s $(pwd)/claude-code-plugin-automations/claude-component-builder/agents/* .claude/agents/
ln -s $(pwd)/claude-code-plugin-automations/claude-component-builder/skills/* .claude/skills/
ln -s $(pwd)/claude-code-plugin-automations/claude-component-builder/commands/* .claude/commands/

ln -s $(pwd)/claude-code-plugin-automations/self-improvement-plugin/agents/* .claude/agents/
ln -s $(pwd)/claude-code-plugin-automations/self-improvement-plugin/skills/* .claude/skills/
ln -s $(pwd)/claude-code-plugin-automations/self-improvement-plugin/commands/* .claude/commands/

Using plugin.json

If your Claude Code version supports it, use the plugin manifest:

# Add to your project's .claude/settings.json
{
  "plugins": [
    "/path/to/claude-code-plugin-automations/claude-component-builder",
    "/path/to/claude-code-plugin-automations/self-improvement-plugin"
  ]
}

🎯 Use Cases

Building Custom Agents

Create specialized agents for your workflow:

  • Code review agents
  • Testing agents
  • Documentation agents
  • Security audit agents
  • Refactoring agents

Creating Auto-Invoked Skills

Build skills that automatically activate:

  • Data analysis expertise
  • Language-specific knowledge
  • Framework-specific guidance
  • Domain-specific best practices

Automating Workflows

Design slash commands for common tasks:

  • Git workflows (commit, push, PR creation)
  • Testing and building
  • Code generation and scaffolding
  • Deployment and releases

Policy Enforcement

Implement hooks for validation:

  • Prevent writes to protected directories
  • Validate bash commands for security
  • Auto-format code after changes
  • Enforce coding standards

📚 Documentation

Quick Links

Claude Code Resources

🏗️ Project Structure

claude-code-plugin-automations/
├── .claude-plugin/
│   └── marketplace.json       # Marketplace manifest
├── claude-component-builder/             # Meta-agent plugin
│   ├── .claude-plugin/
│   │   └── plugin.json        # Plugin manifest
│   ├── agents/                # Meta-architect agent
│   ├── skills/                # 4 builder skills
│   ├── commands/              # 5 creation commands
│   └── README.md
├── self-improvement-plugin/   # Self-improvement plugin
│   ├── .claude-plugin/
│   │   └── plugin.json        # Plugin manifest
│   ├── agents/                # Self-critic agent
│   ├── skills/                # Quality analysis skills
│   ├── commands/              # Review commands
│   ├── hooks/                 # Feedback hooks
│   └── README.md
├── README.md                  # This file
├── LICENSE                    # MIT License
└── .gitignore                 # Git ignore rules

🛍️ Marketplace Compatibility

This repository is fully compatible with Claude Code marketplaces! The .claude-plugin/marketplace.json file enables:

  • Easy Discovery: Browse plugins from Claude Code
  • One-Command Install: Install plugins with claude plugin install <name>
  • Auto-Updates: Keep plugins up-to-date automatically
  • Metadata: Rich plugin information including categories, keywords, and versions
  • Strict Mode: Full validation against plugin manifests

Marketplace Metadata

  • Total Plugins: 2
  • Categories: Development Tools, Productivity
  • All plugins include: Complete manifests, documentation, and examples
  • Compliance: Follows 2025 Claude Code plugin standards

Contributing Improvements Back

The self-improvement plugin creates a meta-feedback loop where it can identify and contribute improvements to itself! See the complete workflow:

📖 Marketplace Contribution Workflow - Detailed guide on how improvements flow from pattern detection → implementation → marketplace distribution

Key Features:

  • 🔄 Self-discovery of plugin limitations through pattern analysis
  • 🛠️ Structured improvement and testing workflow
  • 📦 Version management and semantic versioning
  • 🚀 Automated distribution through marketplace
  • 📊 Impact tracking and continuous measurement

This creates a true continuous improvement cycle where the plugins improve themselves based on usage patterns!

🔧 Development

CI/CD Pipeline

This repository includes a comprehensive CI/CD pipeline for quality assurance:

Automated Validation:

  • Runs on all PRs and commits to main/develop
  • Validates all plugin components (agents, skills, commands, hooks)
  • Enforces conventional commit format
  • Checks JSON schema compliance

Changelog Management:

  • Automatic changelog generation from commits
  • Semantic versioning support
  • GitHub release automation

Commands:

# Validate all plugins
npm run validate

# Quick validation
npm run validate:quick

# Generate changelog
npm run changelog

# Bump version (patch/minor/major)
npm run version:patch
npm run version:minor
npm run version:major

Learn more: CI/CD Guide

Creating Your Own Plugin

Use the Agent Builder to create new plugins:

/claude-component-builder:plugins:new my-awesome-plugin

Claude will guide you through:

  1. Defining the plugin structure
  2. Creating component directories
  3. Writing the manifest file
  4. Adding documentation

Contributing New Plugins

To add a plugin to this repository:

  1. Create your plugin using the Agent Builder
  2. Validate all components using the validation scripts
  3. Document your plugin with a comprehensive README
  4. Test all components thoroughly
  5. Submit a PR with your plugin in a new directory

Plugin Guidelines

All plugins should:

  • Follow Claude Code naming conventions (lowercase-hyphens)
  • Include a plugin.json manifest
  • Provide comprehensive documentation
  • Include validation scripts where applicable
  • Follow security best practices
  • Include usage examples

🧪 Testing

Validate Components

The Agent Builder includes validation scripts:

# Validate an agent
python claude-component-builder/skills/building-agents/scripts/validate-agent.py .claude/agents/my-agent.md

# Validate a skill
python claude-component-builder/skills/building-skills/scripts/validate-skill.py .claude/skills/my-skill/

# Validate a command
python claude-component-builder/skills/building-commands/scripts/validate-command.py .claude/commands/my-command.md

# Validate hooks
python claude-component-builder/skills/building-hooks/scripts/validate-hooks.py .claude/hooks.json

Manual Testing

Test components by:

  1. Creating test cases for agents, skills, and commands
  2. Triggering hooks with various scenarios
  3. Verifying auto-invocation of skills
  4. Checking error handling and edge cases

🤝 Contributing

Contributions are welcome! Here's how you can help:

Report Issues

Found a bug or have a feature request? Open an issue

Submit Plugins

Have a useful plugin to share? Submit a PR with:

  • Complete plugin directory
  • Comprehensive README
  • Validation scripts (if applicable)
  • Usage examples
  • Tests

Improve Documentation

Help improve:

  • Plugin documentation
  • Usage examples
  • Best practices guides
  • Tutorial content

Code Review

Review PRs from other contributors:

  • Check for security issues
  • Verify naming conventions
  • Test functionality
  • Suggest improvements

📋 Naming Conventions

All components follow strict naming conventions:

  • Lowercase letters, numbers, and hyphens only
  • No underscores or special characters
  • Maximum 64 characters

Specific Conventions

  • Plugins: plugin-name (descriptive, unique)
  • Agents: agent-name (action-oriented: code-reviewer, test-runner)
  • Skills: skill-name (gerund form: analyzing-data, generating-reports)
  • Commands: command-name (verb-first: run-tests, create-component)
  • Hooks: Event-based names (descriptive of validation/action)

🔒 Security

Best Practices

When creating plugins:

  • Minimize tool permissions: Start with Read, Grep, Glob
  • Validate all inputs: Check for injection attacks
  • Avoid hardcoded secrets: Use environment variables
  • Review bash commands: Audit for dangerous operations
  • Test security: Try to bypass your own validations

Reporting Security Issues

Found a security vulnerability? Please email security@example.com (or open a private security advisory)

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

🙏 Acknowledgments

  • Anthropic for creating Claude and Claude Code
  • The Claude Code community for inspiration and feedback
  • Contributors who help improve these plugins

🔗 Links

💬 Support

🗺️ Roadmap

Completed Plugins

  • ✅ Agent Builder Plugin: Meta-agent for building Claude Code extensions
  • ✅ Self-Improvement Plugin: Continuous improvement through self-critique and feedback loops

Future Plugin Ideas

  • Testing Suite Plugin: Automated test generation and execution
  • Documentation Plugin: Auto-generate docs from code
  • Security Scanner Plugin: Vulnerability detection and reporting
  • Performance Analyzer Plugin: Code optimization suggestions
  • Git Workflow Plugin: Advanced git operations and automation
  • API Generator Plugin: REST/GraphQL API scaffolding
  • Database Plugin: Schema management and migration tools
  • Code Review Plugin: Automated code review with best practices
  • Refactoring Plugin: Intelligent code refactoring suggestions

Built with ❤️ using Claude Code - A meta-agent building meta-agents! 🤖

Star History

If you find these plugins useful, please consider starring the repository! ⭐

本页内容由上游仓库自动镜像，如有疑问请访问原始仓库。