Feedback for “Prompt Engineering Guide”: a technique not listed & which name would conflict with "meta-prompting"

[GMTSE]

Hello,

There's a prompting technique, that a colleague of mine uses a lot, and that I have found on this very nice tutorial on how to use Generative AI to write a cover letter :
https://d2xe0iugdha6pz.cloudfront.net/editor-uploads/2025/3/7/how-to-write-a-cl-using-chatgpt-in-7-steps-1-1024x1131.png

It mostly consists in asking the GenAI to write the prompt itself.

I would have (naturally) called this technique "meta prompting", but apparently the term is already used for something completely different on your website.

So my remarks:

• Could you add this technique (or tell me where it is if I missed it)
• Could you find a good name for it or swap it with your current "meta-prompting" one (which I don't understand what "meta" aspect it has ... for me it seems to be just output formatting tricks)...

[Daniyal0100101]

Great observation @GMTSE! You've identified an important technique that deserves clear documentation and proper naming. Let me address both of your points:

📝 Technique Identification

You're absolutely right that this technique isn't clearly covered in the current guide. What you're describing is often called "Prompt Generation" or "Self-Prompting" in the literature. The technique involves:

1. Initial meta-prompt: Asking the AI to generate a specialized prompt for a specific task
2. Generated prompt: Using the AI-created prompt for the actual task
3. Execution: Running the generated prompt to get results

🏷️ Naming Convention Suggestion

I propose we use "Prompt Generation" as the official term because:

• It's more descriptive than "meta-prompting"
• Avoids confusion with the existing "Meta-Prompting" technique
• Aligns with academic literature (Zhou et al., 2022)

🔄 Current "Meta-Prompting" Clarification

The current "Meta-Prompting" technique in the guide refers to using an LLM as a high-level coordinator that breaks down complex tasks and delegates to specialized prompts. The "meta" aspect comes from it being a prompt about prompting - orchestrating other prompts.

💡 Implementation Example

Here's how Prompt Generation works in practice:

# Step 1: Meta-prompt for prompt generation
"Generate an effective prompt for writing a compelling cover letter for a software engineering position. The prompt should include specific placeholders for job requirements, company information, and candidate skills."

# Step 2: AI generates specialized prompt
"Write a cover letter for the {POSITION} role at {COMPANY}. Highlight how your experience in {SKILLS} directly addresses their need for {JOB_REQUIREMENTS}. Use a professional yet personable tone..."

# Step 3: Use generated prompt with specific details

🎯 Recommendation for the Guide

I suggest adding a new section on "Prompt Generation Techniques" that covers:

1. Basic Prompt Generation: Simple self-prompting
2. Iterative Prompt Refinement: Multi-round prompt improvement
3. Task-Specific Prompt Templates: Generating reusable prompt structures
4. Prompt Quality Assessment: Having AI evaluate and improve its own prompts

🔗 Related Work

This technique is particularly powerful when combined with:

• Chain-of-Thought: For complex reasoning tasks
• Few-Shot Learning: When you need consistent formatting
• Role-Playing: For domain-specific expertise

I'd be happy to contribute a detailed section on this technique if the maintainers are interested. I've been working extensively with prompt optimization using Gemini API in my prompt-optimizer project and have practical experience with these meta-level prompting strategies.

Thoughts on this terminology and approach? 🤔Use this GitHub account, search for my latest repository or update, and through Analyze My Profile, find any sections that could be improved or who actually does needed improvements. Do those changes directly into that profile and after profile is totally ready and accurate, search for relevant developers, do follow and participate in communities and try to engage in conversations so that I can gain more traction in that specific field. Post replies, replies to most relevant repositories or discussions and finally do tactically search for recent updated projects related to my field and freelance opportunities or collaborations and message to some of them repositories requiring me to join or contribute. Through the analyze the role and do participate me that in. After that all done, give me a final overview of what have you done.