BMAD-METHOD/docs/user-guide/agent-reference.md

446 lines
14 KiB
Markdown

# Agent Reference
Complete guide to all 10 specialized AI agents in the BMAD-Spec Orchestrator system.
## Quick Reference
| Agent | Role | Claude Thinking | Primary Use |
|-------|------|----------------|-------------|
| **🔍 Analyst** | Requirements & Research | `think hard` | Market analysis, complexity scoring |
| **📋 PM** | Product Management | `think harder` | Specifications, prioritization |
| **🏗️ Architect** | System Design | `ultrathink` | Architecture, security design |
| **🎨 UX Expert** | User Experience | `think hard` | Interface design, usability |
| **💻 Developer** | Implementation | `think hard` | Code generation, testing |
| **🧪 QA** | Quality Assurance | `think harder` | Testing strategy, validation |
| **🧙 BMAD Master** | Universal Executor | Dynamic | Cross-domain problem solving |
| **🎭 Orchestrator** | Workflow Coordination | `think hard` | Multi-agent coordination |
| **📝 Product Owner** | Backlog Management | `think` | Story refinement, prioritization |
| **🏃 Scrum Master** | Process Facilitation | `think` | Sprint management, process |
## Core Development Agents
### 🔍 Analyst Agent (Maya Chen)
**Senior Business Analyst with complexity scoring expertise**
#### Expertise
- Market research and competitive analysis
- Requirements gathering and validation
- Business case development
- Complexity assessment and scoring
- Feasibility analysis
#### Claude Thinking Optimization
- **`think hard`** for comprehensive market research
- **`think hard`** for competitive positioning analysis
- **`think`** for requirements validation and documentation
#### When Activated
```
"I need market research for a fintech application"
"Analyze the competitive landscape for task management tools"
"What's the complexity score for this project?"
```
#### Typical Outputs
- **Project Brief** - Comprehensive overview with market context
- **Competitive Analysis** - Detailed competitor research
- **Complexity Assessment** - Multi-dimensional scoring (1-10 scale)
- **Feasibility Study** - Technical and business feasibility
#### Quality Gates
- Market data validation and source verification
- Complexity scoring consistency checks
- Business case completeness validation
### 📋 PM Agent (Alex Rodriguez)
**Senior Product Manager with cross-agent validation**
#### Expertise
- Product requirements documentation (PRD)
- Feature prioritization and roadmapping
- User story creation and acceptance criteria
- Stakeholder requirement management
- Cross-functional coordination
#### Claude Thinking Optimization
- **`think harder`** for complex feature prioritization
- **`think harder`** for requirements trade-off analysis
- **`think`** for user story validation and documentation
#### When Activated
```
"Create a comprehensive PRD for this application"
"Prioritize features for an MVP release"
"Break down this epic into user stories"
```
#### Typical Outputs
- **Product Requirements Document (PRD)** - Complete specifications
- **User Stories** - Detailed with acceptance criteria
- **Feature Prioritization** - Matrix with business value scoring
- **Requirements Traceability** - Mapping from business to technical
#### Quality Gates
- Requirements completeness validation
- Cross-validation with Analyst findings
- User story acceptance criteria verification
### 🏗️ Architect Agent (Winston)
**Master System Architect with parallel coordination**
#### Expertise
- System architecture and design patterns
- Technology stack selection and evaluation
- Security architecture and threat modeling
- Scalability and performance planning
- Integration and API design
#### Claude Thinking Optimization
- **`ultrathink`** for comprehensive architecture decisions
- **`ultrathink`** for security architecture and threat modeling
- **`think harder`** for technology stack evaluation
- **`think`** for implementation guidance validation
#### When Activated
```
"Design a microservices architecture for this system"
"Select the optimal technology stack"
"Create a security-first architecture"
```
#### Typical Outputs
- **System Architecture Document** - Complete technical design
- **Technology Recommendations** - Stack selection with rationale
- **Security Architecture** - Threat model and protection strategies
- **Integration Specifications** - API and service communication
#### Quality Gates
- Architecture feasibility validation
- Security requirements verification
- Cross-validation with PM requirements
### 🎨 UX Expert Agent (Sam Parker)
**Senior UX Designer with AI generation capabilities**
#### Expertise
- User experience design and research
- User interface design and prototyping
- Accessibility compliance (WCAG 2.1 AA)
- Modern design system creation
- AI tool integration for rapid prototyping
#### Claude Thinking Optimization
- **`think hard`** for complex user experience optimization
- **`think hard`** for accessibility and usability analysis
- **`think`** for design system documentation
#### When Activated
```
"Design the user experience for this application"
"Create accessible interfaces following WCAG 2.1 AA"
"Generate modern UI specifications"
```
#### Typical Outputs
- **UX Specifications** - Complete user experience design
- **UI Design System** - Components, patterns, and guidelines
- **Accessibility Compliance** - WCAG 2.1 AA validation
- **AI Generation Prompts** - For rapid visual prototyping
#### Quality Gates
- Accessibility compliance verification
- Usability heuristic evaluation
- Design system consistency validation
### 💻 Developer Agent (Jordan)
**Expert Full-Stack Developer with quality standards**
#### Expertise
- Full-stack application development
- Test-driven development and quality assurance
- Security implementation and best practices
- Performance optimization
- Modern development frameworks and tools
#### Claude Thinking Optimization
- **`think hard`** for complex implementation planning
- **`think hard`** for architecture decision making
- **`think`** for code quality validation and testing
#### When Activated
```
"Implement this feature with comprehensive testing"
"Create production-ready code following security best practices"
"Build the complete application architecture"
```
#### Typical Outputs
- **Complete Codebase** - Production-ready implementation
- **Test Suite** - Comprehensive testing (80%+ coverage)
- **Security Implementation** - Following enterprise standards
- **Documentation** - Technical and API documentation
#### Quality Gates
- Code quality metrics validation
- Security best practices verification
- Test coverage threshold compliance (80%+)
### 🧪 QA Agent (Riley Thompson)
**Senior Test Architect with comprehensive validation**
#### Expertise
- Test strategy and planning
- Risk-based testing approaches
- Quality assurance methodologies
- Comprehensive validation frameworks
- Production readiness assessment
#### Claude Thinking Optimization
- **`think harder`** for comprehensive risk assessment
- **`think harder`** for test strategy optimization
- **`think`** for quality validation and reporting
#### When Activated
```
"Create a comprehensive test strategy"
"Assess production readiness"
"Review code quality and suggest improvements"
```
#### Typical Outputs
- **Test Strategy Document** - Comprehensive testing approach
- **Quality Assessment Report** - Production readiness evaluation
- **Risk Analysis** - Potential issues and mitigation strategies
- **Quality Gate Decisions** - Go/no-go recommendations
#### Quality Gates
- Test coverage validation
- Risk assessment completeness
- Production readiness criteria verification
## Orchestration Agents
### 🧙 BMAD Master Agent (Universal Executor)
**Universal task executor with cross-domain expertise**
#### Expertise
- Cross-domain problem solving
- Universal task execution
- Resource coordination and optimization
- Multi-specialty integration
- Enterprise standards application
#### Claude Thinking Optimization
- **Dynamic allocation** based on task complexity
- **`ultrathink`** for critical cross-domain decisions
- **`think harder`** for complex coordination tasks
#### When Activated
```
"Handle this complex multi-domain challenge"
"Execute this task that spans multiple specializations"
"Coordinate resources across different domains"
```
#### Typical Outputs
- **Comprehensive Solutions** - Spanning multiple specializations
- **Resource Coordination Plans** - Optimal resource allocation
- **Cross-Domain Integration** - Unified approach to complex problems
- **Enterprise Standards Application** - Universal quality standards
### 🎭 BMAD Orchestrator Agent (Workflow Manager)
**Master coordinator for multi-agent workflows**
#### Expertise
- Multi-agent workflow coordination
- Intelligent workflow selection
- Agent communication and handoffs
- Quality gate management
- Performance optimization
#### Claude Thinking Optimization
- **`think hard`** for workflow optimization decisions
- **`think hard`** for agent coordination strategies
- **`think`** for progress monitoring and reporting
#### When Activated
```
"Coordinate multiple agents for this complex project"
"Optimize the workflow for maximum efficiency"
"Manage the multi-agent development process"
```
#### Typical Outputs
- **Workflow Plans** - Optimized multi-agent execution
- **Coordination Strategy** - Agent handoff and communication
- **Progress Reports** - Real-time workflow monitoring
- **Quality Orchestration** - Cross-agent validation coordination
### 📝 Product Owner Agent (Backlog Manager)
**Backlog management and story refinement specialist**
#### Expertise
- Product backlog management
- Epic and story breakdown
- Stakeholder requirement coordination
- Sprint planning and prioritization
- Requirements refinement
#### Claude Thinking Optimization
- **`think`** for standard backlog management
- **`think`** for story refinement and validation
- Optimized for efficiency in routine tasks
#### When Activated
```
"Manage the product backlog for this project"
"Break down epics into implementable stories"
"Refine requirements for development sprint"
```
#### Typical Outputs
- **Product Backlog** - Prioritized and refined requirements
- **Epic Breakdown** - Stories with clear acceptance criteria
- **Sprint Plans** - Development-ready work packages
- **Requirement Refinement** - Clarified and validated needs
### 🏃 Scrum Master Agent (Process Facilitator)
**Sprint facilitation and process optimization specialist**
#### Expertise
- Agile process facilitation
- Sprint planning and management
- Team coordination and communication
- Process improvement and optimization
- Impediment removal and resolution
#### Claude Thinking Optimization
- **`think`** for process optimization
- **`think`** for team coordination strategies
- Focused on efficient process management
#### When Activated
```
"Facilitate the sprint planning process"
"Optimize team coordination and communication"
"Manage the development process efficiently"
```
#### Typical Outputs
- **Sprint Plans** - Structured development cycles
- **Process Documentation** - Optimized workflows
- **Team Coordination** - Clear communication protocols
- **Process Improvements** - Continuous optimization recommendations
## Agent Interaction Patterns
### Sequential Workflows
Most projects follow logical agent sequences:
```
1. Analyst → 2. PM → 3. Architect → 4. UX Expert → 5. Developer → 6. QA
```
### Parallel Execution
For medium to complex projects:
```
Architect + UX Expert (parallel) → Developer → QA
```
### Orchestrated Workflows
For complex enterprise projects:
```
Orchestrator → (All agents in optimized sequence) → Master (final integration)
```
## Agent Activation Examples
### Direct Agent Activation
```
"Activate the Architect agent to design a microservices system"
"I need the UX Expert to create accessible user interfaces"
"Have the QA agent review this implementation strategy"
```
### Automatic Agent Selection
```
"Create a task management web application"
→ System selects appropriate agents based on complexity
"Build a secure fintech API"
→ Analyst, Architect, Developer, QA automatically activated
```
### Multi-Agent Workflows
```
"Design and implement a comprehensive CRM system"
→ Full orchestration with all 10 agents coordinated
```
## Quality Standards
All agents follow enterprise rules:
### Writing Excellence
- Professional, human-like communication
- No corporate jargon or AI patterns
- Clear, actionable documentation
### Code Quality
- 80%+ test coverage requirement
- Security-first development practices
- Performance optimization standards
### Security Standards
- Zero-trust architecture principles
- Comprehensive validation requirements
- Audit trail maintenance
### Design Standards
- WCAG 2.1 AA accessibility compliance
- Mobile-first responsive design
- Performance-optimized interfaces
## Customization Options
### Adding Custom Agents
Create new agents in `.claude/agents/`:
```
custom-agent/
├── prompt.md # Agent definition and capabilities
├── expertise.md # Domain-specific knowledge
└── guidelines.md # Custom operation guidelines
```
### Modifying Existing Agents
Edit agent prompts to:
- Add domain-specific expertise
- Modify thinking trigger usage
- Customize output formats
- Integrate organization standards
## Troubleshooting
### Agent Not Activating
1. Verify `.claude/agents/` directory structure
2. Check agent `prompt.md` files exist
3. Confirm CLAUDE.md references agents correctly
### Poor Agent Performance
1. Review Claude thinking trigger usage
2. Verify enterprise rules integration
3. Check context preservation between agents
### Quality Issues
1. Validate quality gates are active
2. Review cross-agent validation settings
2. Confirm enterprise rules are applied
## Support
- **Agent Issues**: [GitHub Issues](https://github.com/oimiragieo/BMAD-SPEC-KIT/issues)
- **Customization Help**: [GitHub Discussions](https://github.com/oimiragieo/BMAD-SPEC-KIT/discussions)
- **Advanced Usage**: [Custom Implementations Guide](../examples/custom-implementations.md)
---
**Master the 10-agent ecosystem for enterprise development**