Tighten a prompt so the model stops making things up.
AI & PromptingAnalysis💾 0 savesby PromptChief Team
My prompt is producing made-up / wrong info. Help me fix it.
THE PROMPT:
"""
[[PROMPT]]
"""
WHAT IT GETS WRONG: [[PROBLEM]]
Diagnose and patch:
1. Where the prompt invites guessing or overconfidence
2. Add instructions to say "I don't know" when unsure
3. Require it to cite/ground claims in provided context only
4. Add a self-check or confidence step
5. Constrain scope so it can't wander
Return the patched prompt + what you changed and why.