A curated collection of system prompts, custom instructions, and jailbreak prompts for various LLM providers like ChatGPT, Copilot, Claude, and Gemini. This repository offers educational value for users looking to improve their prompt engineering skills and understand LLM security vulnerabilities.

How It Works

The library provides a diverse set of prompts categorized by their function, including system prompts, custom instructions, jailbreak prompts, and prompts designed for protecting LLM instructions. This approach allows users to explore different prompt engineering techniques and learn about potential security risks like prompt injection.

Quick Start & Requirements

• Install: No installation required; access prompts directly from the repository.
• Requirements: Access to various LLM providers (e.g., ChatGPT, Claude, Gemini) to test prompts.
• Resources:
  • A Tale of Reverse Engineering 1001 GPTs
  • Reverse engineering OpenAI's GPTs
  • Understanding and protecting GPTs against instruction leakage
  • GPT-Analyst: A GPT assistant used to study and reverse engineer GPTs

Highlighted Details

• Extensive collection of prompts for popular LLM providers.
• Focus on educational value for prompt engineering and LLM security.
• Includes techniques for extracting system prompts from LLMs.
• Encourages community contributions of new prompts and instructions.

Maintenance & Community

• Open to community contributions for system prompts and custom instructions.
• Links to related projects like GPT-Analyst are provided.

Licensing & Compatibility

• The repository does not explicitly state a license. Users should exercise caution regarding usage and redistribution.

Limitations & Caveats

The repository's content is intended for educational and informational purposes only, and the authors disclaim liability for any misuse. There is no explicit license, which may pose compatibility issues for commercial or closed-source projects.