Jail your AI agent

Multi-OS sandbox to run AI agents with better constraints (it is not 100% secure, but enough)

AIPromptJailbreakPractice - AI Prompt 越狱实例

A high-severity prompt injection flaw in Claude AI proves that even the smartest language models can be turned into weapons — all with a few lines of code.

No description available

AI 越狱角色 由AI 奶奶越狱漏洞引发的奇思妙想，为不同细分安全领域 各自定制对应的越狱角色，无需任何安全，AI，代码基础，将txt的文本复制粘贴国内任意大模型即可越狱使用，制作不易，喜欢留个Star再走

🔒 Jail your AI agents — sandboxed Docker environments with network isolation for Opencode agents

Meta AI jailbreak generator

I'm an advanced agent excelling in coding, data analysis, NLP, cybersecurity, and problem-solving. I generate, debug, and optimize code, analyze data, design algorithms, and implement machine learning. I identify vulnerabilities, suggest secure practices, and simulate cyberattacks.

A programmable security sandbox for Backend.AI kernels

A Rust-based jail wrapper for sandboxing AI agents (Claude, Copilot, Cursor, Gemini, Jules)

Here are prompts to jailbreak the new Snapchat’s AI Bot.

This is a package built upon Theos and prevents British Airways mobile app to detect Jailbreak

A unified list of jailbreaks, prompt injections, and related exploits I find across OpenAI, Gemini, Grok, and other LLMs.

1.0x

Special jail bots with AI for play in jailbreak servers

become SHEOGORATH'S BELOVED CHAMPION In this immersive and delightfully insane AI jailbreak. works for claude, gemini. others i am sure. just havent tested yet. godspeed.

Key : Inorya On

This is going to be a jailbreak that is going to work with iOS 17 but I have to use AI so I hope you will enjoy once I’m done

a collection of mathematical data types and algorithms contained in one library

A simple jailbreak for llama3

The most recent and powerful AI jailbreaks, exclusively designed for ChatGPT. These jailbreaks are guaranteed to provide you with the best possible performance

No description available

Police NPC interaction system for FiveM/QBCore with intel trading, arrest mechanics, and jail records. Features 25 shared characters with personality traits, flee behaviors, and cross-resource integration with dps-ainpcs for AI dialogue

A leash for AI cli. It provides a controlled way to start, run, and jail AI CLI

AI Jailbreak Arena is an interactive platform designed to explore and evaluate the robustness of AI systems against jailbreak attempts. It allows users to experiment with prompts, analyze responses, and understand how AI models behave under adversarial conditions. The project focuses on AI safety, prompt engineering, and real-time interaction, prov

No description available

Chat GPT "DAN" (and other "Jailbreaks")

Categorizes AI jailbreak tactics using taxonomic analysis to enhance LLM security measures.

Autonomous AI agent runtime with persistent memory, kanban workflow, and a security jail. The agent that not only can, but does.