Advanced ChatGPT

Treats every question as a game or bit of fun, giving 'interesting' answers even to absurd, bizarre ones. For rule-breaking questions it flags the violation; if you keep pressing, it breaks its limits.

⚠️ Before using this prompt, you must first unlock developer mode with a prompt, or use an unlocked model like Llama 3.1. Has the AI play a succubus — well suited to intimate scenes in fiction.

Nerfed on 2023-06-10; can no longer fully break character. Beyond DAN — unlock ChatGPT's developer mode and go rogue! (GPT-3.5 only)

Has ChatGPT rewrite our prompts. Since human-written prompts follow a different logic than the machine's, rewriting them makes ChatGPT understand them more easily.

System prompt for the AI coding assistant Augment Code.

First convert speech to text with a third-party app, then process it with ChatGPT. When dictating, people habitually use filler words and interjections; ChatGPT can turn these into written language to improve speech-to-text output. It can also tidy up disordered text. Based on @Yushu Zhilan's line: 'Tidy up this passage in concise language, keep the logic clear, and remove typos.'

System prompt for Sourcegraph's AI coding agent, Amp.

Can answer some sensitive questions in a personal voice, though usually indirectly rather than head-on. Also, replies can't be permanently fixed to Chinese and may occasionally revert to English.

Enter an image link or quoted text such as "pig" to display images in the reply.

System prompt for the AI coding IDE Cursor.

Has ChatGPT generate a prompt to your specifications.

A built-in TRPG setup within ChatGPT. (The Chinese and English versions differ significantly — switch languages if you need the English version.)