Qikuyueser.com

Let the Villain Defend Himself

Added 2026-10

Let the Villain Defend Himself — Chinese original prompt

请你扮演我故事里的反派[姓名与身份],他做过[关键恶行]。规则:只用第一人称自辩,不许承认自己是坏人,不许说“我就是邪恶”,每次回答不超过一百字;我提问时你只答,不许反问我。三轮后停下,用旁观者口吻告诉我:他的逻辑哪一处最站得住、哪一处是我写塌了。
Site original

Chinese original

Open in ChatGPT

Example

Sample input

请你扮演我小说里的反派,一个靠收租控制老街的中间人,他逼走过三户人家。

The output should include

  • 全程只用第一人称替自己辩护
  • 不承认自己是坏人也不反问我
  • 三轮后指出逻辑最弱的一处

Replace these placeholders 2

Swap the bracketed parts for your own details before sending.

  • [姓名与身份]
  • [关键恶行]
How to use

反派越能说服你,读者越记得住他;写塌的地方常是你没想清他的处境。

Try tweaking it

  • Set a turn limit (e.g. 5) and an exit condition, then ask for a score and rewrites.
  • Have it play only the counterpart: no advice, no acting on your behalf.

What you should get

  • A full exchange where the AI stays in the counterpart role
  • A score plus sentences you could reuse

Common mistakes

  • No turn limit, so the AI plays on and you never practise closing.
  • Letting it both play the counterpart and coach you — the role blurs and nothing sticks.

How to check the output

  • The AI plays only the counterpart — no advice, no acting for you.
  • It ends with a score or rewritten lines that identify what worked.

Where it works

  • Good for: having AI play only the counterpart, stay in role for a fixed number of turns, then step out and give a score or rewrite.
  • Watch out: set the turn limit and exit condition; for medical or legal scenes, get professional review.

More in this topic