🧠 How to Be a Better Prompt Engineer
Crafting better prompts to get 10x better results from AI tools
Prompt Engineering is the most underrated skill in 2025
Whether you’re debugging a complex bug, writing documentation, or organizing sprint planning notes, your ability to ask clear, structured prompts directly impacts the quality of answers you get from ChatGPT or other LLMs.
I just created a new digital product “60 AI Prompts to make you a Superstar Engineer”
This is free for all Paid Subscribers. Check out the “Paid Subscribers” page for 100% off coupon
In this newsletter, we will discuss
The anatomy of a good prompt
Use cases and Examples
5 Prompting techniques that work
Common Pitfalls to avoid
FREE digital product for paid subscribers
🛠️ What Is Prompt Engineering?
Prompt engineering is the skill of communicating with large language models (LLMs) in a structured, contextual way to guide them toward giving precise, actionable, and relevant responses.
Think of it as API design for humans—except the "API" is the AI model, and your prompt is the request body.
⚖️ The Anatomy of a Good Prompt
Let’s break this down with side-by-side comparisons:
🧪 Example 1: Code Refactoring
❌ Bad Prompt
"Fix this code."
🟥 Why it's bad: No context, unclear goal, and no guidance on the expected style or formatting.
✅ Better Prompt
You're a senior Python engineer.
Refactor the following function to:
- Follow PEP8 standards
- Improve readability
- Add meaningful comments
Output only the revised function code.
<insert function>
✅ Why it works:
Assigns a role → helps the model frame its response in that mindset
Defines clear goals (PEP8 + readability + comments)
Specifies the output format
🧪 Example 2: Debugging
❌ Bad Prompt
"Why is my code broken?"
🟥 Why it's bad: No code, no error, no reproduction context.
✅ Better Prompt
I’m getting a `NoneType` error when calling this function.
Here’s the traceback and function definition.
Can you help me identify the issue and suggest a fix?
<insert traceback>
<insert code>
✅ Why it works:
Mentions specific error type
Shares exact code and traceback
Asks for diagnosis and fix explicitly