9.9 KiB
🚀 BMAD Quality Framework Enhancement Guide
Master all enhanced features with step-by-step workflows, agent selection guide, and advanced techniques.
🎯 Quick Start Overview
The BMAD Quality Framework transforms basic AI agent orchestration into enterprise-grade quality engineering. This guide provides practical examples and workflows to get you productive immediately.
📋 Step-by-Step Workflows
🔄 Core Development Cycle
1. Start Your Story Development
*develop-story
What happens:
- Systematic story implementation with dual-track progress
- Automatic story file checkbox updates
[x] - Real-time File List maintenance
- TodoWrite tool integration for transparency
Example Output:
✅ Task 1: Create user authentication component [x]
✅ Task 2: Add login validation logic [x]
⏳ Task 3: Implement session management [ ]
📁 Files Modified:
- src/components/AuthComponent.tsx (new)
- src/utils/validation.ts (modified)
- src/types/auth.ts (new)
2. Validate Your Work Quality
*reality-audit
What happens:
- 10-phase comprehensive quality audit
- Automatic A-F grading with composite scoring
- Zero-touch remediation execution
- Automatic options presentation
Example Output:
🔍 Reality Audit Results: Grade B (Score: 82/100)
📊 Breakdown:
- Reality Score: 88/100 (35.2 points)
- Regression Prevention: 78/100 (27.3 points)
- Technical Debt: 80/100 (20.0 points)
✅ Automatic Actions Taken:
- Generated remediation story for TypeScript strict mode
- Split oversized authentication story into 3 focused stories
- Updated architectural compliance documentation
📋 Your Options:
A. [5 min] Fix TypeScript issues and re-audit → Grade A eligible
B. [15 min] Refactor for better pattern compliance → Comprehensive improvement
C. [2 min] Proceed with current quality → Acceptable for development
3. Handle Issues Automatically
When quality issues are detected:
🤖 Automatic Remediation (Zero Commands Required):
- Quality issues → Remediation stories generated automatically
- Oversized stories → Auto-split into manageable pieces
- Mixed concerns → Surgical fix stories created immediately
Example:
🤖 AUTO-REMEDIATION EXECUTED:
Issue Detected: Authentication story contains 12 tasks (>8 limit)
📋 Auto-Generated Stories:
1. "Auth-Core: Login/logout functionality" (4 tasks)
2. "Auth-Validation: Input validation and error handling" (3 tasks)
3. "Auth-Session: Session management and persistence" (5 tasks)
✅ Original story updated with cross-references
✅ New stories added to backlog with proper context
✅ No manual commands required - ready to continue development
🆘 When You Get Stuck
Loop Detection & External Collaboration
After 3 failed attempts on the same issue:
*escalate
What happens:
- Automatic generation of collaboration prompts
- Copy-paste ready for external LLMs (Gemini, GPT-4, Claude)
- Context preservation for seamless handoffs
Example Output:
🔄 LOOP DETECTED: 3 failed attempts on TypeScript integration
🤝 COLLABORATION PROMPT GENERATED:
Copy this to Gemini/GPT-4/Claude for additional perspective:
---
COLLABORATION REQUEST: TypeScript Integration Challenge
CONTEXT: Working on authentication system integration with strict TypeScript requirements.
FAILED ATTEMPTS:
1. Generic type definitions - caused circular dependencies
2. Interface-based approach - broke existing validation logic
3. Utility type solution - increased complexity without solving core issue
SPECIFIC CHALLENGE: Need to integrate new auth types with existing validation system without breaking current implementations or introducing circular dependencies.
CONSTRAINTS:
- Must maintain backward compatibility
- Cannot modify existing validation interfaces
- TypeScript strict mode required
QUESTION: What alternative TypeScript patterns could solve this integration challenge while maintaining clean architecture?
---
💡 This prompt contains all necessary context for effective collaboration.
🎛️ Agent Selection Guide
👨💻 Development Agents
James (Developer Agent)
When to use: Implementation, debugging, code reviews Optimized for: Precision and technical accuracy (temp=0.4)
*develop-story # Systematic implementation
*reality-audit # Quality validation
*build-context # Pre-fix investigation
*escalate # When stuck after 3 attempts
Best for:
- Feature implementation
- Bug fixing with context analysis
- Code quality improvements
- Technical debt resolution
Quinn (QA Agent)
When to use: Quality validation, testing, compliance Optimized for: Systematic analysis (temp=0.3)
*reality-audit # Manual quality audit
*audit-validation # Auto-remediation audit
*create-remediation # Generate fix stories
*story-code-audit # Cross-reference implementation
Best for:
- Quality gate enforcement
- Regression testing validation
- Compliance checking
- Remediation story generation
🎨 Creative & Strategic Agents
Business Analyst
When to use: Requirements analysis, stakeholder needs Optimized for: Creative ideation (temp=0.8)
Best for:
- Requirements gathering
- Stakeholder interview analysis
- Business process optimization
- User story refinement
Architect
When to use: System design, technical decisions Optimized for: Balanced creativity (temp=0.6)
Best for:
- System architecture design
- Technology stack decisions
- Integration planning
- Technical roadmap creation
UX Expert
When to use: User interface design, experience optimization Optimized for: High creativity (temp=0.75)
Best for:
- Interface design
- User experience flows
- Accessibility compliance
- Design system creation
🚀 Advanced Techniques
🎯 Quality Score Optimization
Understanding Composite Scoring:
- Reality Score (40%): Actual vs simulated implementation
- Regression Prevention (35%): Consistency with existing patterns
- Technical Debt (25%): Long-term maintainability
Pro Tips for Grade A (90-100):
-
Perfect Reality Score:
- No mock implementations
- Full functional code
- Comprehensive integration testing
-
Excellent Regression Prevention:
- Follow existing architectural patterns
- Maintain API consistency
- Comprehensive backward compatibility
-
Minimal Technical Debt:
- Clear, self-documenting code
- Proper error handling
- Scalable design patterns
🔧 Workspace Collaboration Features
Cross-IDE Session Management
*workspace-init # Initialize collaborative session
*workspace-status # Check collaboration context
*workspace-handoff # Context-aware agent transitions
*workspace-cleanup # Automated maintenance
*workspace-sync # Synchronize context
IDE-Optimized Commands
Auto-detected environments:
- Cursor, Claude Code, Windsurf, Trae, Roo, Cline, Gemini, GitHub Copilot
What happens automatically:
- Uses IDE-native tools instead of bash commands
- Eliminates approval prompts
- Batches CLI operations when needed
- Leverages integrated panels and runners
📊 Token Efficiency Strategies
Smart Resource Management:
- 78-86% token reduction through intelligent routing
- Lightweight operations for routine tasks (300-800 tokens vs 2,000-5,000)
- Session-based caching eliminates repeated analysis
Optimization Techniques:
- Use targeted commands:
*reality-auditvs generic analysis - Leverage caching: Repeated operations use cached results
- Context-aware execution: Complexity matched to task requirements
- Pattern reuse: Proven approaches over custom solutions
🛠️ Troubleshooting Common Scenarios
📋 "What should I do next?"
Solution: Automatic options presentation after every audit
📋 Your Options:
A. [5 min] Quick fix for immediate progress
B. [15 min] Comprehensive improvement
C. [2 min] Proceed with current state
🔄 "Stuck in debugging loop"
Solution: Automatic escalation after 3 failed attempts
- Copy-paste prompts generated automatically
- External LLM collaboration ready
- Context preservation maintained
📈 "Low quality scores"
Solution: Automatic remediation execution
- Issues detected → Fix stories generated automatically
- No manual command requirements
- Zero-touch resolution process
🗂️ "Story getting too complex"
Solution: Automatic story splitting
- Oversized stories (>8 tasks) split automatically
- Surgical focus stories created
- Cross-references maintained
🎯 Success Patterns
✅ Effective Development Flow:
*develop-story→ Systematic implementation*reality-audit→ Quality validation- Automatic remediation → Zero manual fixes
- Grade A achievement → Auto-push eligible
✅ Quality-First Approach:
- Never accept simulation patterns
- Always validate against existing patterns
- Leverage automatic remediation
- Trust the composite scoring system
✅ Collaborative Problem-Solving:
- Let loop detection trigger external collaboration
- Use workspace handoffs for complex scenarios
- Leverage IDE-specific optimizations
- Maintain context across sessions
🎉 Next Steps
🚀 Ready to Transform Your Workflow?
- Start with a simple story:
*develop-story - Experience automatic quality validation:
*reality-audit - See zero-touch remediation in action
- Leverage collaborative features when needed
📚 Additional Resources:
- enhancements.md - Complete feature reference
- README.md - Full BMAD Method documentation
- workspace-utils/docs/ - Cross-IDE setup guides
🎯 Transform from basic agent orchestration to enterprise-grade quality engineering. Every feature designed for systematic accountability, automated workflows, and collaborative intelligence.