Over the past couple of months, I’ve been obsessively studying and experimenting with Prompt Engineering — not just the theory, but the practical systems that consistently generate high-quality outputs from models like ChatGPT, Claude, Gemini, etc.

During this, I realized something important:

-> Most people don’t struggle with AI… they struggle with STRUCTURE.
Once you give the model a clear role, audience, context, constraints, and a proper workflow… the results multiply instantly.

So I ended up creating a full framework-based system for crafting powerful prompts.
Sharing the most useful pieces here so they can help someone else too:

-> The MAGIC Framework

The MAGIC framework is a handy formula to remember the key ingredients of a powerful prompt, especially for conversational Al like ChatGPT. "MAGIC" here is an acronym:

• M – Make it assume a role
• A – Add context
• G – Give it a format
• I – Instruct it clearly (with the first prompt)
• C – Clarify and iterate with follow-ups

Each letter corresponds to a step in writing the prompt. Let's break down each part with an explanation and example:

● M: Make it assume a role. –

Start your prompt by telling the Al to adopt a certain persona or role. This sets a context and often improves the relevance of the response. Example: "You are an experienced career coach." If you were asking for resume advice, having Al set as a career coach means the suggestions will come from that perspective.

● A: Add context.

Provide any background details or specifics about the situation. This could be the content you want analyzed, the problem details, or the scenario. Example: "The user is a recent college graduate with a degree in computer science, applying for software engineer positions." This context lets the Al tailor its response to that situation, rather than giving generic advice.

● G: Give it a format.

Tell the Al how you want the output. Should it be a list, a narrative, a table, an outline, etc.? Maybe even specify sections. Example: "Provide the advice as a numbered list

of recommendations." For the resume, you might say "Output a professional summary followed by 3 bullet-point suggestions." Format instructions make the answer easier to use.

● I: Instruct it clearly with the first prompt.

This is essentially writing the main question or command – clearly and thoroughly. It should be very clear what you want the Al to do. Example: "Review the following resume for weaknesses and suggest improvements." Combined with earlier bits: "You are an experienced career coach (role). I have a resume below (context) … Please evaluate it and then provide a 5-point list of improvements (instruction + format)." The first prompt should aim to get a good answer without needing clarification.

● C: Clarify and iterate with follow-up prompts.

This part is about the process after the initial answer. It reminds you that you might need to clarify or refine. Using MAGIC, you'd expect to possibly ask follow-ups: "Could you elaborate on point 2?" or "Now help me rewrite the summary using those tips." The prompt can even pre-empt this: "If something is unclear, feel free to ask questions. We can refine the prompt." Though you can also just handle it live by reading the answer and asking for tweaks. The key is not to stop at one attempt_ iteration is part of the framework.

Use-Case Example (Resume Writing):

Let's walk through using MAGIC to prompt for resume feedback. Suppose I have a resume text and I want Al's help. Using MAGIC:

● Make it assume a role: I start with. “You are a professional career advisor specializing in tech industry resumes." (Now Al will respond like a career advisor.)

● Add context: "I will provide my resume below. I am a recent computer science graduate with internship experience in web development." (Now it knows the scenario and what to focus on.)

● Give it a format: "Please respond with a brief critique and then a bullet-point list of 5 specific improvements I can make." (Setting how I want the answer structured.)

● Instruct clearly: "Evaluate the resume for any weaknesses or areas of improvement, then suggest how to improve it. Be honest but constructive. (This is the actual ask, clearly stated.)

● Clarify/iterate: I might add, "If you need additional information about my experience or goals, ask me before giving the suggestions." (This explicitly allows iteration, though I could also just wait to see if the Al asks on its own or do follow- ups after.)

Now I would actually provide the resume text (if It's short enough, Inline; if not, I could say it's attached or summaries lt). But for brevity, assume I did include it.

The Al would produce: as a career advisor, in a structured way, 5 bullet points of improvements (maybe "Highlight your programming projects more, Quantify accomplishments, Tailor the objective statement, etc.").

Then I might follow-up: e.g., "Great, could you rewrite my resume's summary statement following those suggestions?" That's the iterate step in action.

This goes on till you achieve your desired output!

——————–

These is just short version, but even these can massively improve the quality of your AI responses.
If you want the full detailed breakdown (From Intro to Advanced) + 50 ready-to-use prompts + complete guide to Advanced Prompting Techniques, I’ve shared the link in the first comment.

Hope this helps someone!
Happy prompting!