BMAD-METHOD/docs

BMAD-Spec Orchestrator documentation

Welcome to the complete documentation for the Claude-native BMAD-Spec Orchestrator system. This enterprise-grade AI orchestration platform transforms simple requests into production-ready software through 10 specialized agents working with Claude Code.

Documentation structure

📚 User guides

• Getting started - Set up and create your first project in 10 minutes
• Agent reference - Complete guide to all 10 AI agents
• Workflow guide - Understanding adaptive workflows and complexity scoring
• Templates guide - Using the 11 document templates effectively

🔧 API documentation

• Claude Code integration - Complete Claude Code setup and activation
• Agent API - How agents communicate and share context
• Template API - Template variable system and customization

🚀 Deployment

• Installation - Step-by-step setup for any project
• Configuration - Advanced system configuration options
• Troubleshooting - Common issues and solutions

💡 Examples

• Basic projects - Simple applications to get started
• Enterprise projects - Complex systems and workflows
• Custom implementations - Advanced customization patterns

Quick navigation

New to BMAD-Spec Orchestrator

1. Installation guide - Copy the .claude folder to your project
2. Getting started - Ask Claude to build your first app
3. Basic projects - Learn through hands-on examples

Experienced users

• Agent reference - Deep dive into all 10 agents
• Advanced configuration - Customize for your needs
• Enterprise examples - Complex workflow patterns

System administrators

• Installation guide - Enterprise deployment patterns
• Configuration - Security and performance tuning
• Claude Code integration - Complete setup reference

How it works

The BMAD-Spec Orchestrator uses 10 specialized AI agents working through Claude Code to transform natural language requests into complete software solutions.

Core agents (6)

• 🔍 Analyst - Requirements and market research
• 📋 PM - Product management and specifications
• 🏗️ Architect - System design with security-first approach
• 🎨 UX Expert - User experience and interface design
• 💻 Developer - Full-stack development and testing
• 🧪 QA - Comprehensive quality assurance

Orchestration agents (4)

• 🧙 BMAD Master - Universal task executor
• 🎭 BMAD Orchestrator - Workflow coordination
• 📝 Product Owner - Backlog and story management
• 🏃 Scrum Master - Process facilitation

Intelligence features

• Claude thinking optimization - Uses ultrathink, think harder, think hard for optimal decision quality
• Complexity scoring - Automatically selects appropriate workflows (1-10 scale)
• Context preservation - Maintains perfect context across all agent interactions
• Quality gates - Comprehensive validation at every step

Key concepts

Agent activation

Agents activate automatically when you describe what you want to build:

"I need to create a web application for project management"

Claude recognizes this as a trigger and activates the orchestration system.

Thinking optimization

Each agent uses Claude-specific thinking triggers for optimal performance:

• ultrathink - Critical decisions (architecture, security)
• think harder - Complex analysis (risk assessment, prioritization)
• think hard` - Important tasks (market research, UX design)
• think - Standard reasoning and validation

Quality standards

All agents follow enterprise rules from .claude/rules/:

• Professional writing without corporate jargon
• Security-first development practices
• 80%+ test coverage requirements
• WCAG 2.1 AA accessibility compliance
• Performance optimization throughout

Enterprise features

Security-first approach

• Never bypasses permissions - No dangerous shortcuts
• Comprehensive validation - Every output verified
• Audit trails - Complete decision logging
• Zero-trust architecture - Security built into every component

Performance optimization

• Parallel execution - UX Expert and Architect run simultaneously
• Intelligent caching - Reduces redundant processing
• Context optimization - Structured data passing between agents
• Resource management - Efficient computational allocation

Quality assurance

• Cross-agent validation - Agents verify each other's work
• Automated quality gates - Every output meets thresholds
• Error recovery - Automatic retry with improved context
• Continuous improvement - System learns from patterns

Support

• Documentation issues - Report problems with these docs
• System issues - Problems with agent execution or quality
• Feature requests - Suggestions for new capabilities
• Community discussions - Share experiences and patterns

Contributing

The system supports customization and extensions:

• Custom agents - Add specialized domain expertise
• Custom workflows - Create patterns for specific use cases
• Custom templates - Adapt documentation to your standards
• Custom rules - Add organization-specific quality standards

See the configuration guide for details on extending the system.

---

Built for enterprise software development with Claude Code