Foundations Lesson 5

Foundations · Lesson 5

Prompt Debugging (Fixing Bad Outputs)

Bad outputs are not random. They usually point to missing instructions. This lesson gives you a simple checklist to diagnose failures and fix prompts fast.

Reading time: ~8 minutes Level: Beginner Goal: Repair process
Previous lesson See upgrades

Learning objectives

  • Recognise why a prompt failed
  • Use a checklist to diagnose missing pieces
  • Repair prompts without starting over
  • Build confidence and consistency

The core idea

When AI output is wrong, it’s usually because one of the basic prompt components was missing or unclear.

Mindset: don’t blame the AI. Ask: “Which instruction did I forget?”

The debugging checklist

Check these in order:

1) Role missing or unclear? Fix: set a clear role (e.g. “You are a beginner-friendly instructor…”).
2) Task too broad? Fix: narrow to one clear job.
3) Constraints missing? Fix: set tone, length, and rules (e.g. “5 bullets, no jargon”).
4) Output format unspecified? Fix: demand structure (“checklist”, “steps”, “table”).
5) Task too complex for one prompt? Fix: switch to step-by-step prompting.

Worked example

Failed prompt

Help me secure my website.

Why it fails

  • No role
  • Task too broad
  • No constraints
  • No output format

Repaired prompt

You are a website security advisor. Explain basic website security to a non-technical beginner. Use a calm, reassuring tone. Limit the answer to 5 bullet points. Avoid jargon. If anything is uncertain, say so clearly.

Same topic. Completely different result.

Quick quiz

Check your understanding. Then you’ve completed Foundations.

1) What does a bad output usually indicate?
2) What should you check first when debugging?
3) When should you switch to step-by-step prompts?
Score: –
Foundations complete ✅ See Advanced & Professional Automation
Get Power Prompt is about clarity, not cleverness.