Your prompt engineering toolkit.
Structured prompt templates, chaining builders, and examples — organized by use case and technique. Copy, paste, adapt.
Prompt Templates
The 5-Part Prompt
The most reliable starting point for any AI task. Covers task, context, audience, tone, and constraints.
Task: [What specifically do you want? Be a verb — write, analyze, compare, summarize, explain] Context: [What is the situation? What background does the AI need?] Audience: [Who will read or use this? What do they already know?] Tone: [How should it sound? Give 2-3 adjectives or an example of the voice you want.] Constraints: [Format, length, must-includes, must-not-includes. If conflicting, say which wins.]
Chain-of-Thought
Ask the AI to reason step by step before answering. Produces significantly better results on analytical problems.
Task: [Your question or problem] Before answering: Identify the key factors, walk through your reasoning step by step, flag what you are uncertain about, then give your answer with your reasoning visible.
Few-Shot Examples
Show the AI 2-3 examples of what you want. Works well for specialized tasks where you need specific format or reasoning.
Here are 3 examples of the output I want: Example 1: [Input] → [Ideal Output] Example 2: [Input] → [Ideal Output] Example 3: [Input] → [Ideal Output] Now produce output for: [Your Input] Match the format and reasoning style shown above.
Role Prompting
Assign the AI a specific expertise and perspective. Shapes depth, tone, and the type of advice given.
You are a [specific professional role, e.g., a senior product manager who specializes in B2B SaaS]. My situation: [describe your context, constraints, and what you are trying to decide or accomplish] Give me advice as this role would give it. Be specific, not generic. Question anything that seems like the wrong approach.
Iterate Until Excellent
The core workflow: write a prompt, evaluate, refine. Repeat until the output is actually good.
Step 1 — Write: [Your first prompt using the 5-part structure] Step 2 — Evaluate: What is good? What is wrong? What is missing? Step 3 — Refine: [Write a second prompt addressing what was wrong. Be specific: "This part was too generic — make it specifically about X."] Step 4 — Repeat: Keep refining until the output meets your standard. Save the final prompt — you will use it again.
Structured Output
Get AI output in exactly the format you need — table, JSON, bullet list, markdown.
[Your main request] Format requirements: [Be specific about the structure you need] If the format cannot be met, explain why and provide the closest alternative. Do not include any preamble or explanation — just the output in the requested format.
Debug a Bad Output
When AI output is wrong or off-target, diagnose before fixing. Most people rewrite from scratch — this is faster.
My AI prompt produced this output: [paste output] What I wanted instead: [describe the gap] What went wrong: [Is it off-topic? Too vague? Wrong format? Missing context?] Rewrite the prompt to fix specifically: [one targeted instruction based on your diagnosis above]
Compare Options
Structured comparison of two or more options with pros, cons, and a recommendation framework.
Compare [Option A] and [Option B] for [my specific situation/context]. For each option: - Key strengths (be specific, not generic) - Key weaknesses or risks - Under what conditions this would be the right choice Then: Given my constraints of [list constraints], which would you recommend and why? What information would change your recommendation?
Prompt Chains
Multi-step workflows where each prompt builds on the last. Run these in sequence for better results than a single-shot prompt.
Research to Draft
What are the 5 most important things to know about [topic] if you are getting started? Be specific and practical, not generic.
Go deeper on [Aspect #N]. Give me specific examples, common mistakes to avoid, and 2-3 tools or frameworks people use for this. Flag anything that is debated or uncertain.
Write a [type — blog post, email, report] about [topic] for [audience]. Use what we covered in the previous exchanges. Tone: [describe]. Include a section on [specific aspect from step 2].
Draft to Polished
Here are my rough notes: [paste notes]. Turn these into a [type] for [audience]. Tone: [tone]. Length: [target].
This is good but: [be specific — needs stronger opening, too wordy in section 2, the conclusion does not follow from the argument, etc.] Rewrite with that fix.
Review this for: (1) clarity — can any sentence be cut or simplified? (2) tone consistency, (3) does the opening hook pull you in? (4) does the closing land? Revise as needed.
Problem Diagnosis
Here is a problem I am facing: [describe in 2-4 sentences]. The specific issue is: [be precise].
Before solving, help me understand what is actually causing this. What are the 3 most likely root causes? How would I verify which one it is?
Based on our discussion, the most likely cause is [X]. Generate 4 approaches to address this. For each: what to do, what could go wrong, and how to know if it worked.
Want to go deeper on prompt engineering?
The Prompt Engineering course covers all of these techniques in depth — with real examples, edge cases, and a 30-day practice plan.