Ai Prompt Debug

Prompt Engineering Debug Lab (Hands-on)

Prompt Engineering Debug Lab

A hands-on mini-lab to practice fixing broken prompts using constraints, examples, and evaluation. Works offline (no API calls). Students write prompts, test them with a checklist, and compare to a reference solution.

Estimated time
30–45 min
Skills practiced
Prompt debugging
Format
4 scenarios

1) Instructions

  1. Pick a scenario from the dropdown.
  2. Read the goal, the broken prompt, and the “failure report”.
  3. Rewrite the prompt to fix the failure. Use the Debug Checklist.
  4. Self-grade using the rubric. Then open the reference solution and compare.
  5. Optional: copy your improved prompt into ChatGPT and see if it produces the desired output.
Tip: Good prompts have role, task, constraints, format, and examples.
Medium Constraints + Format

Scenario 1 — Summarizer that keeps hallucinating

Click “Run Debug Checklist” after writing your prompt.
Reference solution (open after you try)
One strong solution prompt (not the only correct answer):
Rubric (self-grade)