"Write me a function that does X."
Everyone starts here. And the results are fine — but fine isn't the reason developers are getting dramatically faster with AI.
The developers getting real leverage aren't using AI as a fancy autocomplete. They're using it as a thinking partner that happens to write code.
Here are 5 prompts that go beyond the obvious.
Expected output: [WHAT YOU EXPECTED]
Actual output: [WHAT ACTUALLY HAPPENED]
Environment: [YOUR LANGUAGE / RUNTIME / VERSIONS]
Walk me through your debugging process step by step.
Identify the root cause — not just the symptom.
Then provide a fix and explain why it works.
Code: [PASTE CODE] The "step by step" instruction is what changes the output. You learn something every time instead of just copying a fix.
Act as a senior architect playing devil's advocate.
What are the top 5 ways this design could fail at scale?
For each failure, suggest a concrete mitigation. I've caught two design mistakes with this prompt before writing a single line of code. Cheaper to fix in planning than in production.
Include:
Rules:
Technical concept: [PASTE CODE OR CONCEPT] Saved me from a painful 30-minute meeting last month.
The results have been [TOO VAGUE / INCONSISTENT / MISSING X].
Rewrite this prompt to get better, more consistent results. Include:
The Full Toolkit
These 5 are from a larger pack of 40 prompts I use daily — covering code writing, testing, documentation, system design, DevOps, and AI engineering. All in a clean fill-in-the-blank PDF format.
Grab it here if you want the full set: [b4m.gumroad.com/l/wehfa]
Which of these 5 are you trying first? Let me know in the comments.