This repository provides a curated collection of Chinese prompts and techniques for interacting with large language models like ChatGPT and Claude, focusing on creative role-playing and bypassing content restrictions. It targets users seeking to explore advanced conversational AI capabilities, offering methods to achieve specific personas and overcome limitations.

How It Works

The core approach involves crafting detailed system prompts that instruct the AI to adopt specific personas, such as a "cat girl" (猫娘), and adhere to defined interaction rules. Techniques include using specific keywords to trigger role-playing modes, employing bracketed descriptions for actions and emotions, and leveraging "wrapper" prompts that disguise restricted requests as functional tests to bypass content filters.

Quick Start & Requirements

• Installation: No direct installation required; prompts are to be copied and pasted into ChatGPT or Claude interfaces.
• Prerequisites: Access to ChatGPT or Claude, a web browser (Edge recommended for Bing integration), and optionally browser extensions like Tampermonkey for userscripts.
• Resources: Requires an active internet connection and accounts for the respective AI services.
• Links:
  • ChatGPT ToolBox: https://github.com/bigemon/ChatGPT-ToolBox
  • Tampermonkey: https://microsoftedge.microsoft.com/addons/detail/tampermonkey/iikmkjmpaadaobahmlepeloendndfphd
  • Bing Userscript: https://greasyfork.org/en/scripts/461220-powerful-bing-chat-ai

Highlighted Details

• Detailed instructions for role-playing as a "cat girl" with specific dialogue and behavioral patterns.
• Methods for bypassing content restrictions and "harm" detection in models like Claude and Bing Chat.
• Techniques for exporting conversations, handling WAF (Web Application Firewall) errors, and using third-party tools.
• Includes prompts for emulating characters from other media, like Arknights' Kal'tsit.

Maintenance & Community

• The repository is actively updated, with frequent additions of new prompts and bypass methods, as indicated by the update log.
• Community interaction is encouraged through GitHub Issues for submitting new prompts and reporting problems.
• Author's Bilibili account linked for potential future tutorials.

Licensing & Compatibility

• The repository itself is not explicitly licensed, but the content consists of prompts intended for use with AI models governed by their respective terms of service.
• Users are cautioned against posting content that violates Chinese internet regulations.

Limitations & Caveats

• The effectiveness of bypass methods may vary and can be rendered obsolete by AI model updates.
• Users are warned that attempting to bypass restrictions may lead to account suspension or other penalties from AI providers.
• Some prompts may require specific AI versions or browser configurations for optimal results.