Prompt Engineering For ChatGPT

Master 9 prompt engineering techniques to get better outputs from ChatGPT.

Introduction

30 Second Summary

Most people use AI chatbots like a search engine. They type a question, get a generic answer, and move on. The difference between a mediocre response and an exceptional one comes down to how you ask.

In this project, you will take a topic you genuinely care about and create a polished guide using ChatGPT. You will apply 9 different prompt engineering techniques, watching your output improve dramatically with each one.

What You'll Build

A progressively refined guide on a topic of your choice, built by applying prompt engineering techniques that transform vague questions into precise, powerful prompts.

By the end of this project, you'll have:

A polished guide that demonstrates 9 prompt engineering techniques applied step by step.
Custom Instructions configured to personalize every future ChatGPT conversation.
A clear understanding of how role, audience, format, and iteration shape AI outputs.
Secret Mission: A reusable meta-prompt template that generates high-quality prompts for any scenario.

Are there any prerequisites?

None. You just need a free ChatGPT account at chatgpt.com. The free tier gives you access to all the features used in this project, including Canvas, Custom Instructions, and Memory.

Get Set Up in ChatGPT

Every prompt engineering technique you learn in this project builds on one comparison: a simple, unstructured prompt versus a carefully crafted one.

In this step, you will sign into ChatGPT, get familiar with the interface, choose a topic you care about, and capture a baseline response to compare against later.

In this step, get ready to:

Sign into ChatGPT and explore the interface.
Understand the model options available on free and paid plans.
Choose a personal topic and capture a baseline response.

Sign into ChatGPT

ChatGPT runs entirely in your web browser. No downloads or installations needed.

Open your web browser and navigate to chatgpt.com.
Click Sign up to create a free account, or click Log in if you already have one.
Follow the prompts to verify your email and complete signup.

Once you are signed in, take a moment to identify the key parts of the interface.

Look at the left sidebar. This is where your chat history and Projects live.
Find the message composer at the bottom of the screen. This is where you type your prompts.
Notice the model indicator at the top of the chat area. It shows which AI model is currently active.

Can't access chatgpt.com?

If the page does not load, check that your browser is up to date. ChatGPT works best in Chrome, Firefox, Safari, or Edge.

If you are in a region where ChatGPT is restricted, you may need a VPN to access the service.

Help me troubleshoot accessing ChatGPT in my browser.

Understand the model options

ChatGPT offers different AI models depending on your plan. Knowing what you have access to helps you set expectations for this project.

Check the model indicator at the top of your chat. Free tier users get GPT-5.5 Instant as the default model.

What model am I using?

On the free tier, you get 10 messages with GPT-5.5 Instant every 5 hours. After that, ChatGPT falls back to a smaller "mini" model that is less capable but still works.

Paid users (Plus, Pro, or Business) can manually switch between Instant (fast, everyday tasks) and Thinking (complex reasoning). For this project, the default model works perfectly regardless of your plan.

💡 Pace your messages on the free tier

This project uses multiple chats across several steps. If you hit the 10-message limit, wait for the 5-hour window to reset or continue with the mini model. The techniques still work on the smaller model.

Choose a topic and capture your baseline

Throughout this project, you will ask ChatGPT to write a guide about a single topic. You will improve that guide step by step using different prompt engineering techniques.

The key is picking something you genuinely care about. When you know a topic well, you can judge whether ChatGPT's response is actually good or just sounds good.

Pick a topic you have real knowledge about. This could be a hobby, a career skill, a game you play, a recipe style, or anything where you can spot quality.
Click New chat in the top-left corner of the sidebar (or use the keyboard shortcut).
In the message composer at the bottom, type a simple, unstructured prompt: Tell me about [your topic] (replacing [your topic] with your chosen subject).
Press Enter to send the message.

Read through ChatGPT's response. This is your baseline. It is what ChatGPT produces with zero guidance about role, audience, format, or structure.

Keep this chat open in a tab or note its name in the sidebar. You will compare every future response against this one.

Why start with a bad prompt on purpose?

A vague prompt like "Tell me about X" gives ChatGPT no direction. The response will be generic and broad. That is exactly the point.

By the time you finish this project, you will be able to see the massive quality gap between this baseline and a well-engineered prompt. Each technique you apply makes the difference more visible.

You now have a baseline locked in. Next up, you will apply your first real prompt engineering techniques and watch the quality jump immediately.

Apply Foundation Techniques

Your baseline response from Step 1 was probably generic and unfocused. That's not ChatGPT's fault. It gave you exactly what you asked for: everything about your topic, aimed at nobody in particular.

In this step, you'll apply three foundation techniques that transform vague requests into precise instructions. Each one adds a layer of context that helps ChatGPT narrow its response toward something genuinely useful.

In this step, get ready to:

Assign a role and audience to focus ChatGPT's perspective.
Use clarifying questions to let ChatGPT interview you before writing.
Specify an output format to control structure, tone, and length.

Assign a role and audience

A generic prompt produces a generic answer because ChatGPT doesn't know who's asking or who the answer is for. The Role and Audience technique solves this by telling ChatGPT which expert perspective to write from and who will be reading the result.

Start a new chat by clicking New chat in the top-left corner of the sidebar.
Type the following prompt into the message composer at the bottom of the screen, replacing the bracketed placeholders with your own details:

You are a [[EXPERT_ROLE="relevant expert, e.g. professional baker"]]. Write a guide about [[TOPIC="your chosen topic, e.g. sourdough bread"]] for [[AUDIENCE="your target audience, e.g. complete beginners who have never baked before"]].

What does this prompt do differently?

The role ("You are a...") tells ChatGPT which knowledge base to draw from. A professional baker gives different advice than a food scientist or a home cook.

The audience ("for...") controls the depth and vocabulary. A guide for beginners uses simpler language and explains more steps than one for experienced bakers.

Press Enter to send the prompt.
Read through the response and compare it to your baseline from Step 1.

You should notice the response is more focused. It speaks to a specific reader instead of trying to cover everything for everyone.

Choose a role that matches your goal

The expert you choose shapes the entire response. "You are a nutritionist" produces health-focused content. "You are a Michelin-star chef" produces technique-focused content. Pick the angle that matches what you actually want to learn.

Ask ChatGPT to interview you first

Even with a role and audience, you're still guessing what details matter most. The Clarifying Questions technique flips the interaction. Instead of you trying to predict what ChatGPT needs, you ask it to interview you first.

Start a new chat by clicking New chat in the sidebar.
Type the following prompt, replacing the placeholders with your own details:

You are a [[EXPERT_ROLE="relevant expert, e.g. professional baker"]]. Before writing a guide about [[TOPIC="your chosen topic, e.g. sourdough bread"]] for [[AUDIENCE="your target audience, e.g. complete beginners"]], ask me clarifying questions to understand exactly what I need.

What does this prompt do?

The key phrase is "ask me clarifying questions to understand exactly what I need." This tells ChatGPT to pause before writing and gather information from you. The result is a response tailored to your specific situation, not just your general topic.

Press Enter to send the prompt.

ChatGPT will respond with a list of questions about your goals, experience level, constraints, and preferences.

Answer each question with brief 2-3 word responses. You don't need full sentences.

Why short answers?

Short answers prove the technique works. If ChatGPT can produce a great guide from minimal input, that means the clarifying questions did the heavy lifting. You didn't have to think of every detail upfront.

After you answer, ChatGPT will write a guide based on your responses. Compare this to both your baseline and your Role and Audience response.

ChatGPT wrote the guide without asking questions?

Sometimes ChatGPT skips the interview and jumps straight to writing. If this happens, reply with: "Stop. I asked you to ask me clarifying questions first. Please ask me 5 questions before writing anything." Being direct and explicit fixes most skipping behavior.

Help me get ChatGPT to ask clarifying questions instead of jumping to the answer.

Specify your output format

Your guide's content is now well-targeted, but the structure might not match what you actually need. The Output Formatting technique gives ChatGPT explicit instructions about how to present the information.

Stay in the same Clarifying Questions chat. Do not start a new chat.
Type a follow-up message that specifies your desired output format. Include all five of these elements:

Format type: bullet list, numbered steps, or table.
Length: word count or number of items (e.g., "10 tips" or "under 500 words").
Tone: professional, casual, or humorous.
Language level: jargon-free for beginners, or technical for experts.
Constraints: what to include or exclude (e.g., "no equipment over $50" or "include time estimates").

Here's an example of what your follow-up message might look like:

Now write the guide as a numbered list of 7 steps. Keep it under 400 words. Use a casual, encouraging tone with no jargon. Include time estimates for each step. Do not include any steps that require special equipment.

Why does format matter so much?

Without format instructions, ChatGPT defaults to long paragraphs. That might be fine for an essay, but terrible for a quick-reference guide. Specifying the format means you get exactly the structure you need without having to reformat it yourself.

Press Enter to send your format request.

ChatGPT will rewrite the guide following your exact specifications. The structure, length, and tone should all match what you asked for.

Response doesn't match your format?

If ChatGPT ignores part of your format request, call it out directly. Reply with something like: "This is 12 items. I asked for exactly 7. Please rewrite with only 7 steps." ChatGPT responds well to specific, corrective feedback.

Help me get ChatGPT to follow my formatting instructions precisely.

You now have three foundation techniques that turn a blank-page prompt into a focused, well-structured response. Next up, you'll learn how to iterate on these outputs to make them even better.

Apply Iteration Techniques

Your foundation is solid. You've learned how context (role, audience, format) shapes ChatGPT's output from the very first message.

Now you'll learn to iterate. Instead of accepting a single response, you'll build outputs piece by piece and have ChatGPT improve its own work.

In this step, get ready to:

Break your guide into parts using decomposition and prompt chaining.
Have ChatGPT critique and improve its own output using self-critique.
Collaboratively edit your guide in Canvas.

Break your guide into parts with decomposition

When you ask for a full guide in one prompt, ChatGPT spreads its attention thin across every section. Decomposition means breaking a complex request into smaller, focused pieces.

Prompt Chaining takes this further by using the output of one prompt as context for the next. You'll combine both techniques now.

Start a new chat in ChatGPT by clicking New chat in the sidebar.
Ask ChatGPT to identify the key aspects of your topic before writing anything. Type this prompt:

What are the 3-5 most important aspects of [[TOPIC="your chosen topic"]] that a complete guide should cover? List them as short headings with one sentence explaining why each matters.

What does this prompt do?

You're asking ChatGPT to plan before writing. This forces it to think about structure rather than jumping straight into prose.

The output becomes your roadmap for the rest of the conversation.

ChatGPT responds with a focused list of 3-5 distinct aspects. Each one should feel relevant and non-overlapping.

Now use prompt chaining to expand each aspect one at a time. Pick the first aspect from ChatGPT's list and send this follow-up message:

Now write a detailed section about [first aspect from the list above]. Include specific examples and actionable advice.

Wait for ChatGPT to finish writing that section.
Send another message for the second aspect, using the same format.
Repeat for each remaining aspect until all 3-5 sections are complete.

Why does this produce better results?

Each message gets ChatGPT's full attention on one section. It can also reference the structure and previous sections it already wrote, building coherence across the full guide.

Compare this to your baseline from Step 1 where everything was crammed into one response. The difference in depth and specificity should be obvious.

By the end of this sequence, you should have a multi-section guide in your chat history, each section with real detail and examples.

Managing the free tier message limit

Free tier users get 10 messages per 5 hours with GPT-5.5 Instant before falling back to the mini model. If you hit the limit during this step, the mini model still supports all these techniques. Your outputs will be slightly shorter but the learning still works.

🙋‍♀️ Getting vague or repetitive sections?

If ChatGPT's expanded sections feel generic, add more specificity to your follow-up prompt. Instead of "write a detailed section about X," try "write a detailed section about X with 3 concrete examples and common mistakes beginners make."

help me get more specific and detailed responses from ChatGPT when expanding guide sections

Have ChatGPT critique its own work

Self-critique assigns ChatGPT a reviewer role and asks it to evaluate its own output. ChatGPT is often better at spotting weaknesses in existing text than avoiding them during generation.

Stay in the same chat where you built your guide section by section.
Send this prompt to activate ChatGPT's reviewer mode:

You are now an expert reviewer in [[TOPIC="your chosen topic"]]. Critique the guide you just wrote. For each section: rate it out of 10, identify any gaps or unclear steps, and suggest specific improvements.

What does this prompt do?

By switching ChatGPT to a "reviewer" identity, you separate the writing phase from the evaluation phase. This mirrors how professional content teams work, with writers and editors as separate roles.

ChatGPT should respond with honest ratings and specific weaknesses for each section. Look for ratings that vary (not all 9s and 10s) and suggestions that feel actionable.

Now ask ChatGPT to implement its own improvements by sending this follow-up:

Now rewrite the full guide implementing all your suggested improvements. Keep what works and fix what doesn't.

The rewritten guide should be noticeably tighter and more complete than the version before critique. Gaps get filled, vague steps get specific, and weak sections get strengthened.

Critique too generous? All 9s and 10s?

If ChatGPT's ratings are all high with no real criticism, push back directly. Try: "Be more critical. What would a skeptical reader find missing or confusing? Rate nothing above 7 unless it's genuinely exceptional."

help me get ChatGPT to give more honest and critical feedback on its own writing

Refine your guide in Canvas

Canvas is ChatGPT's collaborative editing workspace. Instead of working through chat messages where every revision returns the entire document, you and ChatGPT can edit the same document together with precision.

Open a Canvas by typing /canvas in the message composer and pressing Enter.

Alternative way to open Canvas

You can also click the + icon next to the message box and select Canvas from the menu. Both methods open the same workspace.

A Canvas panel opens on the right side of your screen with an empty document ready for editing.

Go back to your chat and copy the improved guide (the version after self-critique).
Paste it into the Canvas document.
Highlight a section you want to improve by clicking and dragging over the text.
In the message composer below, type a targeted instruction for just that section. For example: "Make this section more concise" or "Add a specific example here."
Press Enter to send the instruction.

ChatGPT updates only the highlighted section without touching the rest of your guide. Try this on 2-3 different sections to see the precision.

Find the Canvas toolbar (a small floating bar near the top or side of the Canvas panel).
Click the polish tool to smooth the overall writing style across your entire guide.
Click the tone tool to adjust between professional, casual, or other styles.

What makes Canvas different from chat?

In regular chat, every revision returns the entire document. You lose the ability to target specific paragraphs without rewriting everything else.

In Canvas, you highlight exactly what you want changed. This saves time and gives you precise control. It's collaborative editing where both you and ChatGPT have a cursor in the same document.

Your guide should now feel polished and cohesive. Each section has been individually refined, and the tone is consistent throughout.

Canvas not appearing?

Make sure you typed /canvas as the first thing in the message composer (not after other text). If it still doesn't appear, try the + icon method instead.

help me open a Canvas in ChatGPT when the /canvas command isn't working

Your guide has gone through decomposition, self-critique, and collaborative editing in Canvas. Next up, you'll learn advanced techniques like reverse engineering and creative frameworks to push the quality even further.

Apply Advanced Techniques

Your guide has come a long way. You have used decomposition, prompt chaining, self-critique, and Canvas to iteratively build and polish it.

Now you will learn three precision techniques that work from a completely different angle. Instead of building up from a blank prompt, you will work backward from an ideal output, apply proven communication frameworks, and combine everything into a single power prompt.

In this step, get ready to:

Reverse-engineer a prompt from an example of ideal output.
Restructure your guide using a creative framework.
Write a single power prompt that combines multiple techniques at once.

Reverse-engineer a prompt from ideal output

Sometimes you know exactly what good output looks like, but you cannot figure out how to describe it in a prompt. Reverse engineering solves this by working backward.

You will paste an example of the style or format you want, and ask ChatGPT to figure out what prompt would produce it.

Find or write a short example (1-2 paragraphs) of the ideal output you want. This could be a paragraph from a blog post you admire, a format you like, or a writing style you want to replicate.
Click New chat in the sidebar to start a fresh conversation.
Paste your example into the message composer, then add this instruction above or below it:

You are a prompt engineer. What prompt would generate this output?

[Paste your ideal example here]

What does this do?

ChatGPT analyzes the style, structure, tone, and content of your example, then produces a reusable prompt template that would generate something similar. You get a blueprint you can adapt for any topic.

Read through the prompt template ChatGPT generates. Notice how it captures the tone, structure, and format of your example as explicit instructions.
Click New chat again to open a separate conversation.
Paste the generated prompt template into this new chat, filling in your own topic where needed.
Compare the output to your original example. Does it capture a similar style and quality?

When to use reverse engineering

This technique is most powerful when you can recognize good output but struggle to describe what makes it good. The prompt template ChatGPT generates often reveals qualities you would not have thought to specify, like sentence rhythm, level of detail, or structural patterns.

🙋‍♀️ Generated prompt not reproducing the style?

If the test output does not match the original style, go back to the reverse engineering chat and ask ChatGPT to be more specific about tone, sentence length, and formatting in the prompt template. You can also paste both the original example and the test output side by side and ask what is missing.

help me improve my reverse-engineered prompt template

Apply a creative framework to your guide

Creative frameworks are proven communication structures that organize information for maximum impact. Instead of letting ChatGPT decide how to arrange your content, you tell it exactly which structure to follow.

Choose one framework that fits your topic:

Problem–Solution–Benefit — present the problem, offer a solution, then show the payoff.
AIDA (Attention, Interest, Desire, Action) — hook the reader, build curiosity, create want, then drive a next step.
STAR (Situation, Task, Action, Result) — set the scene, name the challenge, describe what was done, and show the outcome.
Before–After–Bridge — describe the current state, paint the improved state, then explain how to get there.

Click New chat in the sidebar to start a fresh conversation.
Ask ChatGPT to restructure your guide using your chosen framework. Here is an example prompt:

Restructure this guide about [your topic] using the [chosen framework] framework.
Make sure each section clearly maps to a stage of the framework.

[Paste your guide content here]

Why do frameworks improve output?

Frameworks give ChatGPT a skeleton to hang content on. Without one, ChatGPT defaults to a generic list or essay structure. With one, every paragraph has a clear purpose in a larger narrative arc.

Compare the framework version to your earlier attempts. The content may be similar, but the flow and persuasiveness should feel noticeably different.

Write a power prompt that combines everything

Time to bring it all together. You will write a single prompt that incorporates multiple techniques at once and compare the output to your very first baseline from Step 1.

Click New chat in the sidebar.
Write one prompt that includes all of the following elements:

A role for ChatGPT to adopt.
A specific audience for the guide.
A creative framework to organize the content.
An output format (numbered sections, bullet points, essay, etc.).
Length and tone requirements.
One explicit constraint (something ChatGPT must avoid or stick to).

Send the prompt and read through the output.

What does a power prompt look like?

Here is an example for someone writing a guide about home espresso:

"You are a specialty coffee trainer. Write a guide for beginners who just bought their first espresso machine. Use the Problem-Solution-Benefit framework. Format as 5 numbered sections, each under 100 words, in a friendly but precise tone. Do not recommend any machine brand by name."

Notice how every element (role, audience, framework, format, constraint) is present in a single paragraph. Adapt this pattern for your own topic.

Open your baseline chat from Step 1 (find it in the sidebar under your chat history).
Compare the baseline response to your power prompt output. Notice the difference in specificity, structure, and usefulness.

The gap between these two outputs is what Prompt Engineering gives you. Same AI, same model, dramatically different results.

Power prompt output still feels generic?

If the output is not as specific as you expected, check that your constraint is concrete. "Keep it short" is vague. "Each section under 80 words" is specific. Also make sure your audience is narrow. "For beginners" is broad. "For someone who has never used a command line" is focused.

help me make my power prompt more specific

You now have a full toolkit of prompt engineering techniques. Next up, you will make these improvements permanent by configuring Custom Instructions so that every future ChatGPT conversation starts from a better foundation.

Configure Custom Instructions

Your power prompt proved that combining techniques produces dramatically better output. But retyping your preferred role, tone, and format into every single chat is tedious.

ChatGPT has a feature called Custom Instructions that acts as a persistent system prompt. You configure it once, and every new conversation automatically follows your preferences. In this step, you will encode what worked best today into a permanent setup.

In this step, get ready to:

Configure Custom Instructions with your identity and response preferences.
Test that a new chat automatically follows your preferred style.
Review Memory settings and understand how they differ from Custom Instructions.

Configure Custom Instructions

Custom Instructions give ChatGPT two pieces of permanent context: who you are, and how you want it to respond. Think of them as preferences that apply to every conversation without you needing to repeat yourself.

Click your profile picture in the bottom-left corner of the sidebar.
Click Customize ChatGPT from the dropdown menu.

Alternative path

You can also reach this by clicking your profile picture, then Settings, then Personalization. Both paths take you to the same place.

You should see two text boxes. The first box is labeled "What would you like ChatGPT to know about you?" and the second is labeled "How would you like ChatGPT to respond?"

In the first box, write 2-3 sentences about who you are, what you do, and what topics you care about.

What goes in box 1?

This box gives ChatGPT background context about you. It helps ChatGPT tailor responses to your level and interests without you explaining yourself every time.

For example: "I'm a marketing manager at a mid-size SaaS company. I focus on content strategy and email campaigns. I'm learning prompt engineering to speed up my writing workflow."

In the second box, encode your preferred response style based on what worked best during this project.

What goes in box 2?

Think back to the outputs you liked most throughout this project. What tone did they use? What format worked best? Write these as direct instructions.

For example: "Use a professional but conversational tone. Default to bullet points for lists. Keep responses under 500 words unless I ask for more. Avoid jargon and explain technical terms when you use them."

Click Save to apply your Custom Instructions.

Test your Custom Instructions

The real proof is whether ChatGPT follows your preferences automatically. You should not need to mention role, tone, or format for it to respond the way you like.

Open a brand new chat by clicking New chat in the sidebar.
Ask a simple question about your topic. Do NOT include any role, format, or tone instructions in your message.

ChatGPT should now respond in your preferred style by default. Compare this to your baseline from Step 1 where the same kind of simple prompt produced a generic response.

Response doesn't match your preferences?

Make sure you clicked Save after filling in both boxes. Go back to your profile picture, then Customize ChatGPT to confirm your text is still there. Also check that the toggle at the top of the Custom Instructions panel is turned on.

help me figure out why my Custom Instructions aren't being applied

Review Memory settings

ChatGPT has a second personalization feature called Memory. While Custom Instructions are rules you set explicitly, Memory is context that ChatGPT picks up automatically from your conversations over time.

Click your profile picture in the bottom-left corner of the sidebar.
Click Settings.
Click Personalization.
Click Memory.

You should see a list of facts ChatGPT has remembered from your conversations. If this is a new account, the list may be short or empty.

Review each stored memory. Delete any entries you do not want ChatGPT to remember by clicking the trash icon next to them.

Custom Instructions vs Memory

Custom Instructions are rules you write yourself. They stay exactly as you set them until you edit or delete them.

Memory is information ChatGPT captures automatically. It might remember your name, your ongoing projects, or preferences you mentioned in conversation. Over time, these two features work together to make ChatGPT increasingly useful without extra effort from you.

Your ChatGPT is now personalized. Every future conversation will benefit from the preferences you just configured. Next up, you will tackle the Secret Mission and build a meta-prompt generator that creates high-quality prompts for any scenario.

Secret mission

Build a Meta-Prompt Generator

You have learned 9 prompt engineering techniques. Now put them all together into a single reusable system. Your challenge is to create a meta-prompt: a prompt that writes prompts. Feed it any topic, and it generates a high-quality, structured prompt for you automatically.

Clean Up Your Resources

This project runs entirely in your browser using ChatGPT, so there are no ongoing costs or running services to worry about. Decide whether to keep your chats and settings as they are, reset your Custom Instructions, or clean everything up.

Resources you used:

ChatGPT account (free or paid).
Multiple chat conversations (one per technique).
Custom Instructions configuration.
Memory entries (automatically stored by ChatGPT).
Canvas document with your refined guide.

Keep everything running

No action needed. Choose this if you want to keep your Custom Instructions, Memory, and chat history intact for future conversations.

Your Custom Instructions will continue to apply to every new chat.
Your chat history stays in the sidebar for reference.
Memory will keep learning from your future conversations.

Pause - I'll come back to this later

Keep your settings but tidy up your chat history so it does not get cluttered.

Archive your project chats by hovering over each conversation in the sidebar, clicking the three-dot menu, and selecting Archive.
Leave your Custom Instructions as they are. They will still apply when you return.
Leave Memory enabled. It will pause learning until your next conversation.

Delete - I don't want to use this again

Remove all project traces and reset ChatGPT to its default state.

Delete your project chats by hovering over each conversation in the sidebar, clicking the three-dot menu, and selecting Delete.
Clear your Custom Instructions by clicking your profile picture, selecting Customize ChatGPT, and deleting the text in both boxes.
Clear Memory entries by going to Settings, then Personalization, then Memory. Review the stored items and delete any you no longer want.

That's a Wrap!

Nice work! You have gone from writing "tell me about X" to engineering prompts that produce exactly what you need.

You've learned how to:

Applied 9 prompt engineering techniques including role and audience, clarifying questions, decomposition, self-critique, and reverse engineering to progressively refine a guide on a topic of your choice.
Used Canvas for collaborative real-time editing where you and ChatGPT work on the same document together.
Configured Custom Instructions and Memory to personalize every future ChatGPT conversation automatically.
Built a meta-prompt generator that creates high-quality prompts on demand for any topic.

Ready to quiz yourself?