agents

Claude Code Agents & Plugins

A collection of specialized agents, plugins, and tooling for Claude Code.

Features

Code Review Agents

Five specialized sub-agents for comprehensive code review:

Command	Agent	Model	Purpose
/code-review <target>	Code Reviewer	sonnet	Code quality, DRY, architecture
/security-review <target>	Security Reviewer	opus	OWASP vulnerabilities, threat analysis
/tech-lead <target>	Tech Lead	opus	Architecture, tech debt, risk assessment
/ux-review <target>	UX Reviewer	haiku	Accessibility, error handling, responsive design
/code-simplify <target>	Code Simplifier	haiku	Reduce complexity, remove dead code
/review-all <target>	All agents	mixed	Run all five in parallel

Plugin Generator

Scaffold new Claude Code plugins interactively:

/plugin-generator:generate [plugin-name]

Creates plugins with:

• Commands (slash commands)
• Skills (auto-invoked capabilities)
• Agents (sub-agent definitions)
• Hooks (event handlers)

Directory Structure

agents/
├── agents/                 # Agent prompt definitions
├── commands/               # Slash commands that invoke agents
├── plugins/                # Claude Code plugins
│   ├── code-review-agents/ # Plugin exposing review commands
│   └── plugin_generator/   # Plugin scaffolding tool
├── settings/               # User settings (LSP, MCP, hooks)
├── .claude-plugin/         # Marketplace registry
├── config.json             # Agent metadata
└── run_agent.py            # CLI helper for agent invocation

Installation

Via Claude Code Plugin Manager

# Install code review agents
claude plugin install code-review-agents@agents

# Install plugin generator
claude plugin install plugin-generator@agents

Manual Installation

Copy to your plugins directory:

cp -r plugins/plugin_generator ~/.claude/plugins/
cp -r plugins/code-review-agents ~/.claude/plugins/

Usage

Running Review Agents

# In a Claude Code session
/code-review ./src/
/security-review ./api/
/review-all ./project/

Creating New Plugins

/plugin-generator:generate my-awesome-plugin

CLI Helper

python run_agent.py --list                    # List available agents
python run_agent.py code_reviewer ./src/      # Get Task config for one agent
python run_agent.py --all ./src/              # Get Task configs for all agents

Adding New Agents

1. Create prompt file in agents/<agent_name>.md:

   # Agent Name
   **Model:** sonnet|opus|haiku
   ## System Prompt
   <prompt content>
   

2. Add entry to config.json under agents

3. Create command in commands/<command-name>.md

Model Selection Guide

• opus: Deep reasoning - security analysis, architecture decisions
• sonnet: Balanced - code review, moderate complexity
• haiku: Fast - simple transformations, focused tasks

Documentation

• Hook Development Guide - Building hook-based plugins

License

MIT