Learn Prompt Engineering & LLM Security From Research
In-depth articles on reverse engineering GPTs, understanding sandbox capabilities, and security findings from professional security research.
Engineering Blueprint
Master Prompt Engineering Across Every LLM Platform
0xebMay 2026
A comprehensive open-source library of system prompts, custom instructions, jailbreaks, and protections for ChatGPT, Claude, Gemini, and 20+ other LLMs—with security techniques and educational articles.
Educational ArticlesCustom InstructionsSystem PromptsJailbreak PromptsInstruction ProtectionsTools & Scripts
Install Command
Run this command to deploy the blueprint to your environment.
What does this do
You're building with LLMs but drowning in scattered prompts across different platforms. You need a single authoritative source that shows both how to write effective prompts AND how to protect them. Plus, if you want to learn from reverse-engineered GPTs, understand sandbox capabilities, or study adversarial techniques, you're hunting through forums and Twitter.
How It Works
Browse curated examples of working system prompts, custom instructions, and jailbreak attempts from 1,500+ real GPTs, organized by LLM provider. Reference articles explain memory mechanics, sandbox contents, and security vulnerabilities. Use protection templates to harden your own GPTs, or study attack patterns to understand risks.
Included Skill Groups
Study 1,500+ Real Custom Instructions From ChatGPT GPTs
Reverse-engineered custom instructions and system prompts from 1,500+ published ChatGPT GPTs, organized by capability type.
Access Current System Prompts From 20+ LLM Providers
Official and reverse-engineered system prompts for OpenAI, Anthropic, Google, Meta, Mistral, xAI, and 15+ other LLM platforms.
Study Real Jailbreak Prompts & LLM Vulnerabilities
Working jailbreak and adversarial prompts from real LLMs, documented for security research and defensive learning.
Defend Your GPT Instructions From Extraction Attacks
40+ tested protection techniques to prevent instruction leaking and jailbreaking—from simple one-liners to sophisticated multi-layer defenses.
Automate GPT Management With Python Scripts & Utilities
Command-line tools and Python scripts for analyzing, managing, and working with ChatGPT custom GPTs and LLM integrations.
Key Features
1,500+ Real Custom GPTs
Reverse-engineered working prompts from ChatGPT's store, organized by capability and provider.
System Prompts for 20+ LLMs
Current system prompts from OpenAI, Anthropic, Google, Meta, Mistral, and 15+ other providers in one place.
Security & Protection Techniques
40+ tested instruction protection patterns—from simple one-liners to sophisticated multi-layer defenses.
Educational Research Articles
Deep dives on GPT reverse engineering, memory mechanics, sandbox inventory, and security findings from REcon 2024.
Jailbreak Examples & Exploits
Real-world adversarial prompts demonstrating vulnerabilities—study them to understand what to protect against.
Key Features
1,500+ Real Custom GPTs
Reverse-engineered working prompts from ChatGPT's store, organized by capability and provider.
System Prompts for 20+ LLMs
Current system prompts from OpenAI, Anthropic, Google, Meta, Mistral, and 15+ other providers in one place.
Security & Protection Techniques
40+ tested instruction protection patterns—from simple one-liners to sophisticated multi-layer defenses.
Educational Research Articles
Deep dives on GPT reverse engineering, memory mechanics, sandbox inventory, and security findings from REcon 2024.
Jailbreak Examples & Exploits
Real-world adversarial prompts demonstrating vulnerabilities—study them to understand what to protect against.
Tools in this Blueprint
About This Blueprint
- License
- GPL-3.0
- Industry
- Technology
- Skills
- 1573 standalone