Troubleshooting: Common Issues and Solutions

Goal: Quickly diagnose and fix the most common prompt engineering failures. These solutions work immediately—no guesswork.

This section addresses the 7 most frequent issues students encounter when building their prompt library. Each problem includes symptoms you'll recognize, the root cause, and an exact fix you can apply in under 2 minutes.

Use this flowchart to identify your issue:

Is the output factually wrong?
├─ Yes → Issue 2 (Hallucination)
└─ No ↓

Is the output too vague/generic?
├─ Yes → Issue 1 (Vague Outputs)
└─ No ↓

Does format vary between runs?
├─ Yes → Issue 3 (Inconsistent Format)
└─ No ↓

Are prompt sections ignored?
├─ Yes → Issue 4 (Ignored Instructions)
└─ No ↓

Do outputs copy example content?
├─ Yes → Issue 5 (Few-Shot Over-Fitting)
└─ No ↓

Is randomness too high/low for task?
├─ Yes → Issue 6 (Temperature Issues)
└─ No ↓

Does it break with real-world inputs?
├─ Yes → Issue 7 (Production Failure)
└─ No → Check common mistakes below

You've mastered troubleshooting when:

You can diagnose issues in less than 1 minute using the flowchart | You fix vague outputs by adding specificity requirements | You prevent hallucination with "NOT STATED" rules | You enforce format consistency with exact templates | You prevent ignored instructions with structured sections | You avoid few-shot over-fitting with diverse examples | You set appropriate temperature for task type | You test prompts with adversarial inputs before production

Time saved: Debugging bad prompts used to take 30+ minutes of trial-and-error. With these systematic fixes, you'll solve 90% of issues in under 2 minutes.