Getting your Trinity Audio player ready…

Phase 1: Initial Prompt Generation

User Says:

Generate a detailed prompt engineering guide. The audience is [TARGET ROLE].

Replace [TARGET ROLE] with your specific audience like “software developers”, “content writers”, “customer service reps”, etc.

LLM Responds:

I'd be happy to create a prompt engineering guide for [TARGET ROLE]. To make this as effective as possible, could you provide me with 3-5 specific examples of:

1. Input scenarios that [TARGET ROLE] would typically encounter
2. The ideal output/response you'd want the AI to provide

This will help me craft a prompt that works perfectly for your specific use case.

Format each example like:
**Input:** [specific scenario]
**Desired Output:** [what the AI should respond with]

User Says:

Here are my examples:

Example 1:
Input: [Your first input scenario]
Output: [Your desired response]

Example 2:
Input: [Your second input scenario]  
Output: [Your desired response]

Example 3:
Input: [Your third input scenario]
Output: [Your desired response]

[Add 2-5 examples total]

Generate a prompt that could have generated these outputs, and include a better set of examples.

LLM Responds:

Based on your examples, here's an optimized prompt for [TARGET ROLE]:

---

# [ROLE-SPECIFIC] AI Assistant Prompt

## Context & Role
You are an expert assistant specialized in helping [TARGET ROLE] with [specific domain/tasks]. Your responses should be tailored to their expertise level and professional needs.

## Core Instructions
[Detailed instructions based on the patterns from user's examples]

## Response Format
[Structure derived from analyzing the user's desired outputs]

## Quality Standards
- [Standard 1 based on examples]
- [Standard 2 based on examples]
- [Standard 3 based on examples]

## Enhanced Examples
[5-7 improved examples that follow the same pattern but cover more scenarios]

---

This prompt should now handle your use cases more effectively. Would you like me to evaluate this prompt next?

---

Phase 2: Prompt Evaluation (New Chat/Session)

User Says:

Generate a detailed prompt evaluation guide. The audience is [SAME TARGET ROLE].

LLM Responds:

I'll create a comprehensive prompt evaluation framework for [TARGET ROLE]. Here's a systematic evaluation guide:

# Prompt Evaluation Framework for [TARGET ROLE]

## Evaluation Criteria (1-10 Scale)

### 1. Clarity & Specificity
- Are instructions unambiguous?
- Is the expected output format clear?
- Are edge cases addressed?

### 2. Role Appropriateness  
- Does it match [TARGET ROLE]'s expertise level?
- Uses appropriate terminology?
- Addresses their specific pain points?

### 3. Completeness
- Covers all necessary aspects?
- Includes sufficient examples?
- Provides quality standards?

### 4. Effectiveness
- Likely to produce desired outputs?
- Handles various input types?
- Scalable across use cases?

### 5. Usability
- Easy to implement?
- Appropriate length?
- Clear structure?

## Evaluation Questions
- What works well in this prompt?
- What could be improved?
- Are there any gaps or ambiguities?
- How could it be more effective for [TARGET ROLE]?

Ready to evaluate a specific prompt - please paste it below.

User Says:

Evaluate this prompt:

[PASTE THE ENTIRE PROMPT FROM PHASE 1]

LLM Responds:

# Prompt Evaluation Report

## Overall Score: [X.X]/10

## Strengths ✅
- **[Strength 1]**: [Specific analysis]
- **[Strength 2]**: [Specific analysis]  
- **[Strength 3]**: [Specific analysis]

## Areas for Improvement ⚠️
- **[Issue 1]**: [Specific problem and impact]
- **[Issue 2]**: [Specific problem and impact]
- **[Issue 3]**: [Specific problem and impact]

## Detailed Analysis

### Clarity & Specificity: [Score]/10
[Analysis of how clear and specific the prompt is]

### Role Appropriateness: [Score]/10  
[Analysis of how well it fits the target role]

### Completeness: [Score]/10
[Analysis of whether all necessary elements are included]

### Effectiveness: [Score]/10
[Analysis of likely performance]

### Usability: [Score]/10
[Analysis of ease of use]

## Recommendations
1. [Specific improvement suggestion]
2. [Specific improvement suggestion]
3. [Specific improvement suggestion]

Ready for the next step? I can generate 3 improved alternative versions of this prompt.

User Says:

Generate 3 improved alternative prompts.

LLM Responds:

# Three Improved Prompt Alternatives

## Version A: Streamlined & Direct (Score: [X.X]/10)
**Best for:** Quick, consistent outputs with minimal complexity

[IMPROVED PROMPT VERSION 1 - more concise and direct]

**Key Improvements:**
- Simplified structure
- Clearer action items
- Reduced cognitive load

---

## Version B: Comprehensive & Detailed (Score: [X.X]/10)  
**Best for:** Complex scenarios requiring thorough analysis

[IMPROVED PROMPT VERSION 2 - more detailed and comprehensive]

**Key Improvements:**
- Enhanced context handling
- More robust examples
- Better edge case coverage

---

## Version C: Adaptive & Flexible (Score: [X.X]/10)
**Best for:** Variable input types and dynamic responses

[IMPROVED PROMPT VERSION 3 - more adaptive and context-aware]

**Key Improvements:**
- Dynamic response scaling
- Better context awareness
- Improved flexibility

## Comparison Summary
| Aspect | Version A | Version B | Version C |
|--------|-----------|-----------|-----------|
| Speed | High | Medium | Medium |
| Thoroughness | Medium | High | High |
| Flexibility | Low | Medium | High |
| Complexity | Low | High | Medium |

Which version best fits your specific needs for [TARGET ROLE]?

---

Phase 3: Final Selection & Refinement

User Says:

I choose Version [A/B/C]. Can you refine it further based on my feedback?

[Optional: Add specific feedback or modifications needed]

LLM Responds:

Excellent choice! Version [X] is ideal for your needs because [reasoning].

Here's the refined final version incorporating your feedback:

---

# FINAL OPTIMIZED PROMPT

[REFINED VERSION OF CHOSEN PROMPT WITH USER FEEDBACK INCORPORATED]

---

## What Makes This Version Better:
- [Improvement 1 with explanation]
- [Improvement 2 with explanation]  
- [Improvement 3 with explanation]

## Usage Tips for [TARGET ROLE]:
- [Tip 1]
- [Tip 2]
- [Tip 3]

## Testing Recommendation:
Try this prompt with your original examples to verify it produces the desired outputs. If you need any adjustments, just let me know!

Would you like me to help you optimize any other prompts using this same methodology?

---

Quick Reference Template

For Future Prompt Engineering Sessions:

Step 1: “Generate a detailed prompt engineering guide. The audience is [ROLE].” Step 2: Provide 3-5 input/output examples Step 3: Request prompt generation with improved examples Step 4: NEW CHAT → “Generate a detailed prompt evaluation guide. The audience is [SAME ROLE].” Step 5: Paste your prompt for evaluation
Step 6: Request 3 improved alternatives Step 7: Choose best version and request final refinements

Pro Tips:

• Always use the same target role across all phases
• Be specific with your examples – quality in = quality out
• Don’t skip the evaluation phase – it catches important issues
• Test your final prompt with real scenarios before deploying
• Keep the methodology consistent for best results