All templates
⚡ Advanced Techniques·Works on: claude, chatgpt, gemini
Negative Example Prompting
Technique: Negative Few-Shot
Show the AI exactly what NOT to do. Often more useful than positive examples.
Advanced#negative-prompting#anti-patterns#constraints
0/2
Fill the template
2 placeholders left.
~196tokens
Live preview· click any pill to jump to its field
Use Negative-Example Prompting to constrain the output space. What we want: Anti-patterns (things that LOOK plausible but are WRONG): Protocol: **Step 1 — Acknowledge the anti-patterns.** Restate each anti-pattern in your own words to confirm understanding. Add one example for each that you would have generated but won't. **Step 2 — Generate.** Produce 10 candidates that satisfy the desired-output spec AND avoid every anti-pattern. Number them. **Step 3 — Self-check.** For each generated candidate, verify against each anti-pattern. If any candidate accidentally violates one, remove it and generate a replacement. **Step 4 — Rank.** Pick the top 3 with one-line reasons. Return ONLY the final filtered + ranked top 3.
Inputs0 of 2
›See the lazy version this template replaces
Before — the lazy prompt
Generate <thing>.
Why it works
- Negative examples teach the boundary of the acceptable space; positive examples only teach the centroid.
- Step 1's acknowledgment + 'example I would have generated' surfaces the model's default trap.
- Step 3's self-check is the only way to catch sneaky anti-pattern violations.
Make this one yours
Replace the bracketed placeholders, then paste into the Prompt Fixer to lint your customisation before hitting send.