Claude Code Extensions

A comprehensive collection of production-ready agents, hooks, commands, and output styles for Claude Code

Supercharge your Claude Code experience with 60+ specialized agents, 18 skills, 8 lifecycle hooks, 11 slash commands, and powerful automation tools—all ready to install and customize.

✨ Highlights

• 🤖 60+ Expert Agents - From code reviewers to deep research specialists (React, Django, Temporal, Cloudflare, Home Assistant, and more)
• 🎯 Model-Invoked Skills - Capabilities Claude discovers and uses automatically based on context
• 🔒 Safety-First Hooks - Block dangerous commands, protect sensitive files, audit all operations
• ⚡ Auto-Linting - Integrated Python, Go, and JS/TS linters that run automatically
• 🎯 Smart Orchestration - Tech-lead agents that coordinate multi-step tasks across specialists
• 📦 Easy Installation - Interactive CLI installer with settings.json merging
• 🧪 Battle-Tested - Production patterns from real-world Claude Code workflows

📦 Available Plugins

The repository now provides 21 self-contained plugins for Claude Code marketplace installs:

Install only what you need - no need to load 50+ agents if you only work with React!

🚀 Quick Start

Installation

Option 1: Plugin (Recommended for Multiple Projects)

Install as modular plugins via Claude Code's plugin system - pick only what you need for your tech stack:

# In any Claude Code session:
/plugin marketplace add github:nodnarbnitram/claude-code-extensions

Then install the plugins you need:

# Essential foundation (everyone needs this)
/plugin install cce-core@cce-marketplace

# Add plugins for your tech stack:
/plugin install cce-kubernetes@cce-marketplace    # For Kubernetes development
/plugin install cce-cloudflare@cce-marketplace    # For Cloudflare Workers/AI
/plugin install cce-esphome@cce-marketplace       # For ESPHome IoT
/plugin install cce-web-react@cce-marketplace     # For React/Next.js/TanStack
/plugin install cce-django@cce-marketplace        # For Django backend development
/plugin install cce-grafana@cce-marketplace       # For Grafana plugin development & billing
/plugin install cce-vite@cce-marketplace          # For Vite 8 config/build guidance and Vitest v4 testing
/plugin install cce-typescript-v6@cce-marketplace # For TypeScript 6 migration and tsconfig guidance (skill-only, separate from cce-typescript)
/plugin install cce-research@cce-marketplace      # For deep research tasks

Each marketplace package under plugins/* is now self-contained with local agents/, commands/, skills/, and hooks/ content where applicable.

Benefits:

• ✅ Modular - install only what you need
• ✅ Automatic updates across all projects
• ✅ Centralized management
• ✅ No repository cloning needed

Command namespaces:

• Core: /cce:git-commit, /cce:prime
• Kubernetes: /cce-kubernetes:health
• Linear: /cce-linear:triage, /cce-linear:create-linear-pr
• (Other plugins currently provide agents/skills only)

Option 2: Standalone (For Single Project or Custom Setup)

git clone https://github.com/nodnarbnitram/claude-code-extensions.git
cd claude-code-extensions
./install_extensions.py install ~/my-project

The installer lets you:

• ✅ Choose extensions interactively by type or category
• ✅ Preview changes with --dry-run before applying
• ✅ Merge settings.json safely without conflicts
• ✅ Copy dependencies automatically

# Common usage patterns
./install_extensions.py list                    # See all available extensions
./install_extensions.py info code-reviewer      # Get details on specific extension
./install_extensions.py install --dry-run ~/my-project   # Preview what will be installed
./install_extensions.py install --type agent --category core ~/my-project  # Install specific category

Option 3: Manual Copy

# Copy specific items manually
cp .claude/agents/core/code-reviewer.md ~/my-project/.claude/agents/
cp -r .claude/hooks ~/my-project/.claude/
cp .claude/settings.json ~/my-project/.claude/

Which Installation Method Should I Use?

Creating New Extensions

Agents - Use the built-in meta-agent:

> Use the meta-agent to create a [domain] specialist agent

Hooks - Use Python with uv script format (see CLAUDE.md for templates)

Commands - Create .md files in .claude/commands/ with frontmatter

Output Styles - Create .md files in .claude/output-styles/ with custom prompts

📖 Full guide: See CLAUDE.md for detailed patterns and examples

💡 Why Use This?

For Teams

• Share consistent workflows across your organization
• Enforce code quality standards automatically
• Onboard developers faster with specialized agents
• Version control your Claude Code configuration

For Solo Developers

• Automate repetitive tasks with hooks and commands
• Get expert-level assistance across multiple frameworks
• Keep your main Claude context focused with subagents
• Protect yourself from mistakes with safety hooks

For Learning

• Explore production-ready extension patterns
• Learn from 50+ agent examples across different domains
• Understand hook lifecycle and automation
• See how to structure complex AI workflows

📚 Extension Reference

Commands

Skills

Model-invoked capabilities that Claude automatically discovers and uses based on task context.

Creating new skills:

• Use the skill-creator agent: > Use the skill-creator to create a skill for [purpose]
• Or copy the skeleton template: cp -r templates/skill-skeleton .claude/skills/my-skill

See docs/claude-code/agent-skills.md for detailed guidance.

Agents

Hooks

Python hooks using uv run --script for zero-config dependency management.

Utilities: LLM integrations (Anthropic, OpenAI, Ollama), TTS engines (pyttsx3, OpenAI, ElevenLabs)

📖 Documentation

• CLAUDE.md - Complete guide for creating extensions
• CONTRIBUTING.md - How to contribute to this project
• docs/claude-code/ - Official Claude Code documentation

📋 Requirements

• Claude Code CLI
• uv - Python package manager
• Python 3.11+

🎯 Usage Examples

Auto Code Review

> Review my recent changes

The code-reviewer agent automatically analyzes your git diff.

Desktop Notifications Configure hooks for desktop alerts when Claude needs input (see CLAUDE.md).

Test Hooks Locally

echo '{"tool_name": "Bash", "tool_input": {"command": "ls"}}' > test.json
uv run ./.claude/hooks/pre_tool_use.py < test.json
echo $?  # 0 = allowed, 2 = blocked

🤝 Contributing

We welcome contributions! See CONTRIBUTING.md for:

• Development setup
• Testing guidelines
• Pull request process
• Style guidelines

Quick tips:

• Use the meta-agent to create new agents
• Follow existing patterns in .claude/hooks/
• Test thoroughly before submitting
• Update documentation with your changes

📦 Project Structure

.claude/
├── agents/           # 50+ specialized AI assistants
├── skills/           # Model-invoked capabilities
├── hooks/            # Lifecycle automation scripts
├── commands/         # Reusable slash commands
├── output-styles/    # Custom system prompts
└── settings.json     # Hook configuration

templates/
└── skill-skeleton/   # Starter template for new skills

docs/                 # Claude Code official docs
install_extensions.py # Interactive installer CLI

🔗 Resources

• Claude Code Docs
• Claude Code GitHub
• Report Issues

📄 License

MIT License - Free for everyone, including commercial use.

What this means:

• ✅ Use it anywhere - personal projects, startups, Fortune 500 companies
• ✅ Modify it however you want
• ✅ No permission needed
• ✅ Just keep the copyright notice

That's it. Use it, learn from it, build amazing things. 🚀