@cloudkinetix/bmad-enhanced/expansion-packs/ck-llm-agent-dev/data/prompt-patterns.md

Cloud-Kinetix enhanced fork of BMAD-METHOD - Breakthrough Method of Agile AI-driven Development with robust versioning and unified validation.

# Prompt Engineering Research Guide

Dynamic guide for discovering and applying current prompt engineering patterns based on real-time research.

## Research-First Approach

**IMPORTANT**: Instead of using static patterns, always research current prompt engineering techniques before implementation:

1. **Research Current Patterns**: Use web search to discover latest prompt engineering techniques, frameworks, and best practices
2. **Analyze Use Case**: Understand the specific requirements, constraints, and goals for the prompting task
3. **Validate Approaches**: Research validation methods and testing frameworks for the chosen patterns
4. **Adapt and Optimize**: Customize patterns based on research findings and specific project needs

## Research Areas for Pattern Discovery

### Core Pattern Categories

Research these pattern types based on your specific use case:

- **Role-based prompting** - Research current approaches to persona definition and role specification
- **Chain-of-thought reasoning** - Research latest CoT techniques and reasoning frameworks
- **Few-shot learning** - Research example selection and optimization strategies
- **Structured output** - Research current formatting and constraint techniques

### Advanced Pattern Categories

Research these advanced techniques based on current literature:

- **Constitutional AI** - Research current approaches to value alignment and safety constraints
- **Meta-prompting** - Research prompt generation and self-improvement techniques
- **Multi-modal prompting** - Research cross-modal reasoning and integration patterns
- **Tool-calling patterns** - Research current function calling and API integration approaches

### Safety and Alignment Patterns

Research current safety techniques:

- **Content filtering** - Research current moderation and safety constraint approaches
- **Bias mitigation** - Research current bias detection and prevention techniques
- **Prompt injection defense** - Research current security patterns and defenses

## Research Methodology for Pattern Selection

### 1. Context Analysis

Before selecting patterns, research:

- **Domain-specific best practices** - Look for patterns proven in your specific domain
- **Model capabilities** - Research what works best with your target LLM(s)
- **Performance requirements** - Research patterns that meet your latency/cost constraints
- **Quality standards** - Research validation approaches for your quality requirements

### 2. Current Literature Review

Always research:

- **Recent papers** - Search for latest prompt engineering research and publications
- **Industry case studies** - Look for real-world implementations and lessons learned
- **Community best practices** - Research current community consensus and evolving techniques
- **Tool documentation** - Review latest documentation for prompting frameworks and tools

### 3. Testing and Validation

Research current approaches to:

- **A/B testing** - Methods for comparing prompt variants
- **Evaluation metrics** - Current approaches to measuring prompt effectiveness
- **Automated testing** - Tools and frameworks for systematic prompt evaluation
- **Human evaluation** - Best practices for human-in-the-loop validation

## Dynamic Pattern Generation

Instead of using static patterns, research and generate patterns by:

1. **Researching current techniques** - Find the latest approaches for your specific use case
2. **Analyzing successful implementations** - Study real-world examples and case studies
3. **Adapting to context** - Customize patterns based on your specific requirements
4. **Testing and iterating** - Use research-backed testing methodologies to optimize
5. **Staying current** - Continuously research new developments and improvements

## Implementation Guidelines

- **Start with research** - Always begin by researching current best practices
- **Validate through testing** - Use research-backed evaluation methods
- **Document rationale** - Record why specific patterns were chosen based on research
- **Iterate based on evidence** - Use data and research to guide improvements
- **Stay adaptive** - Regularly research new developments and update approaches

---

**Remember**: This guide emphasizes research-driven pattern discovery over static templates. Always research current best practices and adapt to your specific context and requirements.