Design a prompt engineering system for: [YOUR AI APPLICATION]

**Application Context:**
- Use case: [what the AI does]
- Model: [Claude/GPT-4/etc.]
- Volume: [requests per day]
- Quality requirements: [accuracy, consistency needs]

Create a system that includes:

1. **Prompt Architecture**
   - System prompt structure
   - Dynamic context injection patterns
   - Few-shot example management
   - Output format specifications

2. **Version Control**
   - Prompt versioning strategy
   - A/B testing framework
   - Rollback procedures
   - Changelog format

3. **Testing Framework**
   - Unit tests for prompt behavior
   - Golden test sets
   - Regression detection
   - Quality metrics (accuracy, latency, cost)

4. **Optimization Patterns**
   - Token usage optimization
   - Caching strategies
   - Batching approaches
   - Model routing (different models for different tasks)

5. **Monitoring & Observability**
   - Logging structure
   - Quality dashboards
   - Drift detection
   - Cost tracking

6. **Implementation Template**
   - Folder structure
   - Configuration format
   - Deployment pipeline

Output as a production-ready architecture document.