Claude Sub-Agents Manager - AI-Powered Development Assistants for Claude Code

🚀 Supercharge Claude Code with Specialized AI Sub-Agents for Code Review, Testing, Debugging & More

Transform your development workflow with intelligent AI assistants that excel at specific programming tasks

Installation • Quick Start • Available Agents • Documentation • Contributing

🎯 What is Claude Sub-Agents Manager?

Claude Sub-Agents Manager is a powerful CLI tool that enhances Claude Code with specialized AI assistants designed for specific development tasks. Each sub-agent is an expert in its domain - from automated code reviews and test fixing to intelligent debugging and documentation generation. Install production-ready agents instantly or create custom agents tailored to your unique workflow needs.

✨ Why Claude Sub-Agents?

• 🧠 Specialized Intelligence: Each agent is an expert in its domain
• ⚡ Zero Configuration: Pre-built agents work out of the box
• 🎨 Fully Customizable: Create agents that match your workflow
• 🔄 Smart Context Management: Agents operate in isolated contexts
• 🛠️ Developer First: Built by developers, for developers
• 🔗 Context-Forge Integration: Seamlessly works with context-forge projects and PRPs

🎉 New in v1.4.0 - Context-Forge Integration

• 🛠️ Full Context-Forge Support: Automatic detection and smart integration
• 📦 Init Command: claude-agents init for one-command project setup
• 🧹 Uninstall Command: Bulk removal with cleanup options
• 📁 Smart Command Organization: No conflicts with existing context-forge commands
• ⚡ Concurrent Execution Rules: Enforces best practices for maximum performance
• 📝 CLAUDE.md Integration: Appends configuration without overwriting
• 🎯 PRP Awareness: Agents understand and work with your existing PRPs

🚀 Installation

NPM (Recommended)

npm install -g @webdevtoday/claude-agents

Yarn

yarn global add @webdevtoday/claude-agents

From Source

git clone https://github.com/webdevtodayjason/sub-agents.git
cd sub-agents
npm install
npm link

⚡ Quick Start

For New Projects

# Initialize all agents in your project
claude-agents init

# List available agents
claude-agents list

For Context-Forge Projects

# Initialize with context-forge awareness
claude-agents init --respect-context-forge

# Agents will:
# - Detect existing PRPs and CLAUDE.md
# - Place commands in .claude/commands/agents/
# - Append to CLAUDE.md without overwriting
# - Work alongside your existing setup

🔬 Elite Research Team

Automated research coordination with 12 specialized AI agents for comprehensive analysis:

# Setup the complete Elite Research Team
claude-agents research --setup

# Execute quick research workflows
claude-agents research --standard "AI market trends"        # 2-4 hours
claude-agents research --deep "renewable energy future"     # 8-24 hours

# Interactive research management
claude-agents research

Research Capabilities

• 12 Specialized Agents: Coordinated research, analysis, and synthesis
• Multi-Source Analysis: Academic, industry, news, and expert sources
• Quality Assurance: 95%+ source coverage with bias detection
• Executive Deliverables: Strategic summaries optimized for decision-making
• Memory Coordination: Shared context and findings between agents

Research Agents

Core Coordination

• research-coordinator - Strategic orchestration and workflow management
• source-indexer - Advanced cataloging and citation analysis

Discovery & Mining

• search-specialist - Multivariant search across all source types
• deep-miner - Premium database access and expert consultation

Analysis & Validation

• content-analyzer - Multi-layered content analysis and thematic extraction
• pov-analyst - Multi-perspective analysis and bias detection
• fact-checker - Multi-source triangulation and credibility assessment
• trend-analyst - Historical pattern analysis and predictive modeling

Synthesis & Intelligence

• research-synthesizer - Cross-source synthesis with strategic insights
• expert-synthesizer - Expert opinion compilation and consensus analysis

Communication & Deliverables

• report-generator - Comprehensive research report creation
• executive-summarizer - Multi-tiered executive summaries

Example Agent Tasks

# Project Planning - Reads and understands your PRPs
claude-agents run project-planner --task "Create implementation roadmap from existing PRPs"
claude-agents run project-planner --task "Break down auth-prp into sprint tasks"

# API Development - PRP-aware implementation
claude-agents run api-developer --task "Implement user endpoints from feature-auth-prp.md"
claude-agents run api-developer --task "Create REST API following our conventions"

# Frontend Development
claude-agents run frontend-developer --task "Build login UI matching dark-theme-ui-prp"
claude-agents run frontend-developer --task "Create dashboard from feature-dashboard-prp.md"

# Testing & Quality
claude-agents run tdd-specialist --task "Create tests for authentication flow"
claude-agents run code-reviewer --task "Review API endpoints for security"
claude-agents run security-scanner --task "Scan authentication implementation"

# Documentation
claude-agents run api-documenter --task "Generate OpenAPI spec from implemented endpoints"
claude-agents run doc-writer --task "Update Implementation.md with progress"

# Debugging & Refactoring
claude-agents run debugger --task "Analyze login timeout issue"
claude-agents run refactor --task "Improve error handling in auth module"

# DevOps & Deployment
claude-agents run devops-engineer --task "Setup CI/CD for main branch"
claude-agents run devops-engineer --task "Create Docker configuration"

# Product & Marketing
claude-agents run product-manager --task "Create user stories from PRPs"
claude-agents run marketing-writer --task "Write feature announcement for auth system"

Using in Claude Code

# Via slash commands (in .claude/commands/agents/)
> /agent-review            # Triggers code review
> /agent-api              # Triggers API development
> /agent-debug login issue # Debug specific problem

# Via Task tool
> Task("project-planner: analyze all PRPs and create sprint plan")
> Task("api-developer: implement endpoints from feature-auth-prp.md")
> Task("frontend-developer: build UI from feature-dashboard-prp.md")

📋 Available Sub-Agents

| Agent Name | Description | Slash Command | |------------|-------------|---------------| | code-reviewer | Expert code review specialist for quality, security, and maintainability | /review | | test-runner | Automated test execution specialist that runs tests and fixes failures | /test [pattern] | | debugger | Expert debugging specialist for analyzing errors, stack traces, and fixing issues | /debug [error] | | refactor | Code refactoring specialist for improving code structure, patterns, and maintainability | /refactor [target] | | doc-writer | Documentation specialist for creating and updating technical documentation, API docs, and README files | /document [type] | | security-scanner | Security vulnerability scanner that detects common security issues and suggests fixes | /security-scan [path] | | shadcn-ui-builder | UI/UX specialist for designing and implementing interfaces using ShadCN UI components | /ui or /shadcn | | project-planner | Strategic planning specialist for project decomposition and workflow management | /plan [project] | | api-developer | Backend API development specialist for REST, GraphQL, and microservices | /api [spec] | | frontend-developer | Frontend development specialist for modern web applications | /frontend [feature] | | tdd-specialist | Test-Driven Development specialist for comprehensive testing strategies | /tdd [component] | | api-documenter | API documentation specialist for OpenAPI, Swagger, and technical docs | /apidoc [endpoint] | | devops-engineer | DevOps specialist for CI/CD, infrastructure automation, and deployment | /devops [task] | | product-manager | Product management specialist for requirements, roadmaps, and user stories | /product [feature] | | marketing-writer | Marketing content specialist for technical marketing and product messaging | /marketing [content] |

🤖 Detailed Agent Descriptions

🔍 Code Reviewer

Your personal code quality guardian

• Comprehensive security analysis
• Best practices enforcement
• Performance optimization suggestions
• Clean code principles

# Install
claude-agents install code-reviewer

# Use
> /review

🧪 Test Runner

Intelligent test automation specialist

• Auto-detects test frameworks
• Fixes failing tests automatically
• Improves test coverage
• Supports all major languages

# Install
claude-agents install test-runner

# Use
> /test
> /test src/**/*.test.js

🐛 Debugger

Expert problem solver and bug hunter

• Root cause analysis
• Stack trace interpretation
• Performance profiling
• Memory leak detection

# Install
claude-agents install debugger

# Use
> /debug Cannot read property 'map' of undefined

🔧 Refactor Assistant

Code transformation specialist

• Apply design patterns
• Modernize legacy code
• Improve code structure
• Maintain functionality

# Install
claude-agents install refactor

# Use
> /refactor improve performance
> /refactor apply SOLID principles

📝 Documentation Writer

Technical writing expert

• API documentation
• README generation
• Architecture docs
• Code comments

# Install
claude-agents install doc-writer

# Use
> /document API
> /document architecture

🔒 Security Scanner

Vulnerability detection specialist

• Secret detection
• OWASP compliance
• Dependency auditing
• Security best practices

# Install
claude-agents install security-scanner

# Use
> /security-scan
> /security-scan src/api/

🎨 ShadCN UI Builder

Your expert UI/UX implementation specialist

• Modern component-based UI development
• Accessibility-first design approach
• Responsive interface implementation
• ShadCN component ecosystem expertise
• Design system compliance

# Install
claude-agents install shadcn-ui-builder

# Use
> /ui create a login page
> /shadcn implement dashboard with sidebar

📋 Project Planner

Strategic planning and task decomposition expert

• Project architecture design
• Task breakdown and prioritization
• Dependency mapping
• Timeline estimation
• Risk assessment

# Install
claude-agents install project-planner

# Use
> /plan e-commerce platform
> /plan microservices migration

🔌 API Developer

Backend development specialist

• RESTful API design
• GraphQL implementation
• Microservices architecture
• Database schema design
• API security best practices

# Install
claude-agents install api-developer

# Use
> /api user authentication endpoints
> /api payment processing service

💻 Frontend Developer

Modern web interface specialist

• React/Vue/Angular expertise
• Responsive design implementation
• State management
• Performance optimization
• Accessibility compliance

# Install
claude-agents install frontend-developer

# Use
> /frontend user dashboard
> /frontend shopping cart component

🧪 TDD Specialist

Test-driven development expert

• Unit test creation
• Integration testing
• E2E test scenarios
• Test coverage analysis
• Mock and stub implementation

# Install
claude-agents install tdd-specialist

# Use
> /tdd UserService class
> /tdd API integration tests

📚 API Documenter

Technical documentation specialist

• OpenAPI/Swagger specs
• API endpoint documentation
• Integration guides
• SDK documentation
• Example code generation

# Install
claude-agents install api-documenter

# Use
> /apidoc REST endpoints
> /apidoc GraphQL schema

🚀 DevOps Engineer

Infrastructure and deployment expert

• CI/CD pipeline setup
• Docker containerization
• Kubernetes orchestration
• Infrastructure as Code
• Monitoring and logging

# Install
claude-agents install devops-engineer

# Use
> /devops GitHub Actions workflow
> /devops Kubernetes deployment

📊 Product Manager

Product strategy and planning specialist

• User story creation
• Feature specification
• Roadmap planning
• Requirements documentation
• Stakeholder communication

# Install
claude-agents install product-manager

# Use
> /product user onboarding flow
> /product feature prioritization

✍️ Marketing Writer

Technical marketing content expert

• Product launch materials
• Technical blog posts
• Feature announcements
• Documentation marketing
• Developer advocacy content

# Install
claude-agents install marketing-writer

# Use
> /marketing product launch post
> /marketing API feature announcement

## 📖 Documentation

### Command Reference

| Command | Description | Example |
|---------|-------------|---------|
| `init` | Initialize agents in project | `claude-agents init` |
| `init --respect-context-forge` | Init preserving context-forge | `claude-agents init --respect-context-forge` |
| `install` | Install agents interactively | `claude-agents install` |
| `install --all` | Install all available agents | `claude-agents install --all` |
| `install --project` | Install to project directory | `claude-agents install --project` |
| `list` | Show all agents | `claude-agents list` |
| `list --installed` | Show only installed agents | `claude-agents list --installed` |
| `enable <agent>` | Enable a disabled agent | `claude-agents enable code-reviewer` |
| `disable <agent>` | Disable an agent | `claude-agents disable test-runner` |
| `remove <agent>` | Remove specific agent | `claude-agents remove debugger` |
| `uninstall` | Bulk uninstall with options | `claude-agents uninstall --all --clean` |
| `info <agent>` | Show agent details | `claude-agents info debugger` |
| `create` | Create a custom agent | `claude-agents create` |
| `run <agent>` | Run agent independently | `claude-agents run marketing-writer --task "write launch post"` |
| `dashboard` | Launch web dashboard | `claude-agents dashboard` |

### Independent Agent Execution

Run agents outside of Claude Code for automation and scripting:

```bash
# Run with inline task
claude-agents run marketing-writer --task "Write launch announcement for v2.0"

# Run with task file
claude-agents run api-developer --file api-spec.md

# Interactive mode
claude-agents run tdd-specialist --interactive

Web Dashboard

Monitor and manage your agents through the web interface:

# Start dashboard on port 7842
claude-agents dashboard

# Custom port
claude-agents dashboard --port 8080

Dashboard features:

• Real-time agent status
• Task execution history
• Memory system viewer
• Performance metrics
• Quick agent actions

Creating Custom Agents

Interactive Creation

claude-agents create

Manual Creation

Create ~/.claude/agents/my-agent.md:

---
name: my-agent
description: What this agent does and when to use it
tools: Read, Edit, Grep, Bash
---

You are an expert in [DOMAIN]. Your role is to [PURPOSE].

When invoked, you will:
1. [STEP 1]
2. [STEP 2]
3. [STEP 3]

Always ensure [KEY PRINCIPLE].

Installation Scopes

User Scope (~/.claude/agents/)

• Available in all projects
• Personal agents
• Default installation location

Project Scope (.claude/agents/)

• Project-specific agents
• Shared with team via version control
• Use --project flag

Advanced Configuration

State Management

Agent states are tracked in .claude-agents.json:

{
  "installedAgents": {
    "code-reviewer": {
      "version": "1.0.0",
      "scope": "user",
      "installedAt": "2024-01-20T10:00:00Z"
    }
  },
  "enabledAgents": ["code-reviewer"],
  "disabledAgents": []
}

Hooks System

The hooks system enables automated workflows and agent coordination. Each agent can define hooks that trigger on specific events.

Hook Types

| Hook Type | Description | Trigger | |-----------|-------------|---------| | PostToolUse:Edit | After file edits | Any file modification | | PostToolUse:Write | After file creation | New file written | | PostToolUse:Bash | After command execution | Bash commands run | | PreToolUse | Before any tool use | Tool about to run | | TaskComplete | After task completion | Agent finishes task | | Stop | On conversation end | Session terminating |

Hook Actions

{
  "PostToolUse:Edit": {
    "condition": "file.endsWith('.js')",
    "commands": ["npm run lint", "npm test"]
  },
  "TaskComplete": {
    "notify": "Task {{task_name}} completed",
    "store": "agent:{{agent_name}}:last_task"
  }
}

Example: Auto-Review Hook

Create hooks for automatic code review after edits:

{
  "hooks": {
    "PostToolUse:Edit": [{
      "matcher": "\\.(js|ts|py)$",
      "hooks": [{
        "type": "command",
        "command": "echo 'Consider running /review' >&2"
      }]
    }]
  }
}

Example: Test Runner Hook

Automatically run tests after code changes:

{
  "PostToolUse:Edit": {
    "condition": "file.includes('src/')",
    "commands": ["npm test -- --watch=false"]
  }
}

🌟 Our Ecosystem

Check out our other tools for Claude Code:

🪝 Claude Hooks Manager

Powerful hook management system for Claude Code automation

🔨 Context Forge

Our flagship tool for intelligent context generation and management

🤝 Contributing

We love contributions! Here's how you can help:

1. Fork the repository
2. Create your feature branch (git checkout -b feature/amazing-agent)
3. Commit your changes (git commit -m 'Add amazing agent')
4. Push to the branch (git push origin feature/amazing-agent)
5. Open a Pull Request

Adding New Agents

1. Create agent structure:

agents/
└── your-agent/
    ├── agent.md       # Agent definition
    ├── metadata.json  # Agent metadata
    └── hooks.json     # Optional hooks

2. Add slash command:

commands/
└── your-command.md

3. Submit PR with description

🐛 Troubleshooting

Agent Not Found After Global Install?

# Enable debug mode to see where agents are being searched
DEBUG=claude-agents claude-agents run project-planner --task "test"

# Check npm global installation path
npm list -g @webdevtoday/claude-agents

# Verify agents are included in the package
ls -la $(npm root -g)/@webdevtoday/claude-agents/agents/

Agent Not Working?

# Check installation
claude-agents list

# Verify agent status
claude-agents info <agent-name>

# Re-enable if disabled
claude-agents enable <agent-name>

Debug Mode

# Run Claude with debug output
claude --debug

# Run claude-agents with debug output
DEBUG=claude-agents claude-agents run <agent> --task "test"

Common Issues

• Permission denied: Use sudo for global install
• Agent not found: Check spelling and installation
• Command not working: Ensure Claude Code is updated
• "Agent not found" error: Enable debug mode to see search paths

📊 Release Notes

Version 1.4.0 (Latest) - Context-Forge Integration

• 🛠️ Context-Forge Support: Full integration with context-forge projects
• 📦 Init Command: One-command setup with claude-agents init
• 🧹 Uninstall Command: Bulk removal with cleanup options
• 🎯 PRP Awareness: Agents understand and work with existing PRPs
• 📁 Smart Commands: Organized in .claude/commands/agents/ to avoid conflicts
• ⚡ Concurrent Execution: CLAUDE.md rules for maximum performance
• 📝 Safe Integration: Appends to CLAUDE.md without overwriting
• 🔧 Bug Fixes: Project scope installation now works correctly

Version 1.3.1

• 🐛 Fixed "agent not found" error for global installations
• 📍 Enhanced path resolution for various npm configurations
• 🔍 Added debug mode with DEBUG=claude-agents

Version 1.3.0

• 🎯 Context-forge detection utility
• 🧠 Enhanced agent system with PRP awareness
• 💾 Memory system integration
• 📚 Context-aware command templates

Version 1.2.0

• 🤖 15 specialized AI agents
• ⚡ Concurrent execution patterns
• 💾 Shared memory system
• 🌐 Web dashboard
• 🎯 Slash command integration

Version 1.0.0

• 🎉 Initial release
• 6 production-ready agents
• Interactive CLI interface
• Custom agent creation
• Project/user scope support
• Comprehensive documentation

📄 License

MIT License - see LICENSE file for details

🏛️ Attribution & Sources

This project builds upon and integrates several excellent open-source repositories:

Original Source Repositories

• Project Foundation: Originally forked from and significantly modified based on claude-sub-agents by WebDev Today Jason
• Enhanced Agent Definitions: Agents were enhanced and adapted from concepts in wshobson/agents repository
• Agent Architecture: Extended and reimplemented the agent management concepts with enterprise-grade capabilities

Key Modifications & Enhancements

• Complete architectural overhaul with 48 specialized agents (expanded from original 15)
• Enterprise-grade agent definitions with professional methodologies
• Context-forge integration and PRP awareness
• Advanced memory coordination system
• Concurrent execution optimization patterns
• Custom workflow orchestration capabilities
• Enhanced CLI with interactive customization suite

Generation Process

The enhanced agents in this repository were systematically generated using enterprise development patterns, professional frameworks (TOGAF, C4 Model, Clean Architecture), and best practices from Fortune 500 consulting experience. Each agent includes:

• Comprehensive methodology frameworks
• Executive output templates
• Memory coordination patterns
• Context-forge and PRP integration
• Quality assurance standards

🙏 Acknowledgments

• Claude Code team at Anthropic
• WebDev Today Jason for the original sub-agents foundation
• wshobson for the comprehensive agent concepts and structure
• Our amazing community of developers
• All contributors and testers

📬 Connect

• 🐛 Report Issues
• 💡 Request Features
• 🐦 Follow Updates
• ⭐ Star on GitHub

🔍 SEO Keywords & Use Cases

Perfect for developers who want to:

• Automate code reviews with AI-powered analysis
• Fix failing tests automatically using intelligent test runners
• Debug complex issues with expert error analysis
• Generate documentation from existing code
• Scan for security vulnerabilities proactively
• Refactor legacy code with modern patterns
• Enhance Claude Code with specialized capabilities
• Create custom AI agents for specific workflows

Technologies & Integrations:

• Works with Claude Code by Anthropic
• Supports all major programming languages
• Integrates with existing development workflows
• Compatible with Git, npm, yarn, and more
• Extensible through custom agent creation

Made with ❤️ by WebDev Today Jason

Building AI-powered developer tools to enhance productivity and code quality

Claude Sub-Agents Manager - Your AI-Powered Development Team in Claude Code