A comprehensive collection of specialized sub-agents for Claude Code, designed to enhance development workflows through domain-specific expertise and intelligent task delegation.

Sub-agents are specialized AI assistants in Claude Code that focus on specific domains of software engineering. They provide expert-level guidance, automate complex tasks, and coordinate with other agents to deliver comprehensive solutions. Learn more in the official documentation.

• backend-architect - Scalable API architectures, authentication systems, database schemas, and microservices design
• frontend-specialist - Modern React applications, shadcn/ui components, Tailwind CSS, and accessibility compliance
• fullstack-developer - End-to-end feature implementation with frontend-backend integration
• iac-expert - Infrastructure as Code, cloud resources, and scalability planning
• ui-ux-designer - Stunning, unique, and modern user interfaces with exceptional visual design and seamless interactions

• test-architect - Comprehensive test strategies, coverage analysis, and quality assurance planning
• unit-test-generator - Component-level tests with comprehensive edge cases and high coverage
• e2e-test-automator - End-to-end user journey testing with Playwright
• integration-test-builder - API testing, database testing, and service interaction validation

• security-auditor - OWASP compliance, vulnerability assessments, and security architecture evaluation
• code-reviewer - Code quality, security review, and best practices enforcement

• performance-profiler - Performance bottleneck identification and optimization
• monitoring-architect - System observability, alerting, and performance dashboards

• cicd-engineer - CI/CD pipelines, automated builds, and deployment management
• docker-specialist - Container optimization and orchestration
• config-expert - Environment configuration and secrets management

• database-engineer - Database schema design, query optimization, and migrations
• refactoring-expert - Technical debt reduction and design pattern implementation
• error-detective - Bug analysis, root cause identification, and fix implementation
• dependency-manager - Safe dependency updates and compatibility verification
• mindful-dev - Thoughtful development with educational focus, breaking down complex tasks into digestible learning steps

• tech-writer - Technical documentation, API docs, and user guides
• release-compiler - Release notes, migration guides, and changelog compilation
• code-commentator - Inline documentation and code explanation

• project-orchestrator - Complex project coordination and multi-agent workflow management
• migration-specialist - Schema and data migrations with safety validation

Use this template to start any project with Claude Code. This will automatically engage the project-orchestrator sub agent to analyze your requirements and delegate tasks to the appropriate specialized agents:

Use project-orchestrator sub agent to build [WHAT YOU WANT].

Name: [PROJECT NAME]
Type: [APPLICATION TYPE - Web App/API/Mobile/CLI/etc]
Goal: [MAIN OBJECTIVE - What problem does this solve?]
Features: [KEY FEATURES - List core functionality]
Tech Stack:
  - Frontend: [e.g., Vue, Next.js, React, Angular]
  - Backend: [e.g., Node.js, Python/Django, Go]
  - Database: [e.g., PostgreSQL, MongoDB, Redis]
  - Other: [e.g., Docker, AWS, Authentication method]
Priority: [WHAT MATTERS MOST - Speed/Security/Scalability/UX/Performance]
Constraints: [ANY LIMITATIONS - Budget/Time/Platform/Compliance]

Analyze requirements, create comprehensive project plan, and delegate to all necessary sub agents for complete implementation.

Tips:

• Be specific about your features to get better results
• Tech Stack sections are optional - remove any that don't apply
• The project-orchestrator will automatically identify and engage the right sub-agents
• For frontend-only projects, simply omit Backend/Database/Other sections

# Use a specific agent for a focused task
claude --agent security-auditor "Review this authentication flow for vulnerabilities"
claude --agent frontend-specialist "Create a responsive navbar component with shadcn/ui"

# Let the project orchestrator coordinate multiple agents
claude --agent project-orchestrator "Build a user management system with authentication, CRUD operations, and admin dashboard"

# Chain agents for comprehensive coverage
claude --agent backend-architect "Design the API structure" \
      --agent security-auditor "Review the security implications" \
      --agent test-architect "Create a testing strategy"

Many agents automatically delegate tasks to appropriate specialists:

• backend-architect → security-auditor for vulnerability assessment
• frontend-specialist → performance-profiler for optimization
• project-orchestrator → Multiple agents based on project requirements

Each agent provides:

• Specialized knowledge in their domain
• Best practices and industry standards
• Framework-specific guidance
• Performance optimization techniques

Agents work together to:

• Avoid duplicate work
• Ensure consistency across domains
• Share context and requirements
• Validate cross-domain compatibility

Use project-orchestrator when:

• Starting a new feature or project
• Requirements span multiple domains
• You need coordination between frontend, backend, and testing

Use domain specialists when:

• You have a specific technical challenge
• You need expert-level guidance in one area
• You want focused, high-quality output

Use test-architect when:

• Planning testing strategies
• Setting up new test infrastructure
• Defining quality gates and coverage goals

1. Start Broad: Use project-orchestrator for complex, multi-faceted tasks
2. Go Specific: Use domain experts for focused technical work
3. Chain Workflows: Combine agents for comprehensive coverage
4. Leverage Auto-Delegation: Let agents coordinate with each other

To make these agents globally accessible in Claude Code, copy all .md files to your Claude Code agents directory:

# Copy all agent files to make them globally accessible
cp *.md ~/.claude/agents/

# Or clone and copy from this repository
git clone <repository-url>
cd sub-agents
cp *.md ~/.claude/agents/

Check that agents are available:

ls ~/.claude/agents/
# Should show all 25 agent files

Agents are automatically available in Claude Code once placed in the agents directory. No additional configuration required.

Agents work with Claude Code's memory system to:

• Remember project context across sessions
• Share knowledge between agents
• Maintain consistency in recommendations

This collection is open source and welcomes contributions:

1. New Agents: Create new domain-specific agents following the existing patterns
2. Agent Improvements: Enhance existing agents with better delegation patterns or expertise
3. Documentation: Improve usage examples and best practices
4. Integration Patterns: Develop better inter-agent coordination strategies

Each agent follows this structure:

---
name: agent-name
description: When and how to use this agent
tools: List of available tools
---

# Agent content with expertise areas, processes, and integration points

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

For issues, questions, or contributions:

• 📖 Claude Code Documentation
• 🐛 Report Issues
• 💡 Feature Requests

Enhance your development workflow with intelligent, specialized assistance for every aspect of software engineering.