What it does

This repo is a curated collection of copy-paste prompts designed to make ChatGPT adopt specific personas—Linux terminal, English translator, academic editor, Android interviewer, even a text-based RPG engine. Each prompt is a block of instructions you paste verbatim to constrain the model’s behavior. There’s also a JSON file for programmatic access.

The interesting bit

The “tuning” framing (调教) is bluntly honest about the dynamic: these are leash-and-collar instructions for a probabilistic parrot. The value isn’t technical sophistication—it’s the sheer breadth of social roles someone bothered to formalize, from “front-end expert replacing Google” to “cyberpunk homepage PM.”

Key highlights

• 60k+ stars suggest this filled a real gap for Chinese-speaking users
• Prompts cover ~25 use cases: papers, resumes, ad copy, medical content, children’s books
• JSON export available for integration into other tools
• Includes WeChat groups and a domestic mirror site (aimakex.com)
• README claims a “GPT5.5” based service, which appears to be branding, not a real model version

Caveats

• Heavy promotional content for the mirror site and WeChat channels
• No evaluation of whether prompts actually improve output quality versus simpler instructions
• “GPT5.5” claim is unsupported and likely marketing fluff

Verdict

Grab it if you’re building a Chinese-language prompt library or need quick social-engineering templates for LLMs. Skip if you already write your own system prompts or expect rigorous prompt engineering research.