Learn LLM Prompting With Hundreds of Examples

Developers and security researchers lack a centralized, organized reference for understanding how LLMs are configured, instructed, and jailbroken. Learning prompt engineering requires studying real examples from production systems, and understanding security risks requires seeing actual attack and defense patterns.

1. Browse prompts and instructions by LLM provider (OpenAI, Anthropic, Google, etc.) or by content type (system prompts, custom GPTs, jailbreaks, or protections). 2. Read detailed articles on reverse-engineering GPTs, ChatGPT sandbox capabilities, and memory mechanisms. 3. Study working examples to understand instruction syntax, tone, constraints, and capability design. 4. Use security protection examples to lock down your own GPTs. Output: a hands-on education in how LLMs are actually built and secured.

Anyone from prompt engineers to security researchers can learn from 1,500+ real-world examples instead of starting from zero.

The repository is a documentation and reference library; no execution environment is required. All content is organized in markdown files by LLM provider and content category. Prompts span multiple versions of models (OpenAI has prompts from 2023 to 2025, Anthropic Claude versions 1.0 through 4.5, Google Gemini versions 1.5 through 2.5, etc.), so timestamps and version tags help identify which system each prompt applies to.

This repository is purely educational and informational. Users assume full responsibility for how they apply this material. Jailbreak and security prompt content is documented to educate about risks; the authors explicitly oppose unlawful use and are not liable for misuse. Always respect the terms of service of the LLMs you work with.