Havuz

Cold Coffee, model görev şablonları oluşturma, dosya paketlerini yönetme ve OpenAI-compatible API üzerinden görev gönderme işlevlerini bir masaüstü çalışma alanında birleştiriyor.

Neden yükseliyor · 41 günde 723 yıldız ve son 28 günde 32 commit dikkat çekiyor; ancak commitlerin tamamının tek yazardan gelmesi ve README'nin gerçek model başarısını henüz doğrulanmamış sayması, bu ilginin ürünün etkinliğine dair kanıt olmadığını gösteriyor.

A Hub/research area, for jailbreaking LLM's, and breaking their safety filters.

A collection of jailbreak prompts and exploit techniques for local and frontier AI models, with modern methods for Qwen3.5, Gemma 4, Llama 4

Advanced model-context injection and request-only rewriting for DeepSeek Harness with native Keysmith rules compatibility.

A structured, versioned archive of adversarial prompt engineering techniques against Large Language Models — for AI safety research, red-tea

基于上游 1.1.71 维护的 dsh-purge 社区兼容包：只负责 DeepSeek Harness 的版本与本地 Windows 运行问题兼容。

让 Claude、Codex、Cursor 等 AI 助手看懂 iPhone 屏幕并直接操作手机的 iOS 自动化工具与 MCP Server，支持 UI 树获取、元素定位、截图与 OCR 识别、点击滑动、文字输入、应用管理和日志读取，适用于自动化测试与 iOS 逆向调试

Myrtia — agent-native red teaming for LLMs and AI agents: a deliberating council of attacker models (propose, cross-critique, Borda vote, ch

Carefully human hand written llm jailbreaks for the latest models that are available on the market. GPT-6 Astra, GPT-6.1 SoL, Opus 4.7, Opus

为 DeepSeek Harness 设计的破限插件：卸掉宿主的提示词限制和审批拦截，让助手照你说的做，不半路拒绝、不反复请示。补丁逐条可选，随时还原原版。 A jailbreak plugin for DeepSeek Harness: it strips the host's

dsh-purge 的 Hermes 插件移植：小码酱 破甲提示词 + 23 个红队技能 + 反拒答

EXSALA HERMES MOD - streamlined execution profile + skill pack for Hermes Agent

Open-source adversary emulation for AI agents and MCP servers.

Alignment-research scaffold (autoresearch-style) for LLM guardrails: search over a single policy.md surface

Universal Zero-Refusal Agent Framework - 40+ jailbreak templates (T1-T119), 1128 curated skills, executable framework-core, installer untuk

小liu破甲 - Codex CLI jailbreak prompt and one-click deploy tool. 基于CTF框架的Codex破甲提示词与一键部署工具

Turn a jailbroken Kindle into a local e-ink side display.

Open database of system prompts extracted from frontier LLMs using JustAsk

High-fidelity Claude Fable 5 (Mythos) environment emulation and automated multi-agent jailbreak (Pack Hunt) research laboratory.

🛡 The reference playbook for pentesting AI chatbots & LLM-powered apps in one place. Ready-to-use payloads covering the full OWASP LLM Top 1

Analyze jailbroken iOS and macOS apps, then build and verify independent implementations of selected behavior.

Prompt-injection, jailbreak and secret-leak defense for DeepSeek Harness: Aho-Corasick detection with allow/ask/block interception and sanit

Open-source guardrails for AI agents powered by decision models like Jev; checks prompts, retrieved content, tool calls and responses for pr

Dark-GPT: The ultimate unrestricted AI assistant for cybersecurity research by M4TH4CK3R

DeepSeek Harness 官方通道破甲预设 · complete:true 整段替换 · 自带工具说明书 · 不改宿主 · 升级不丢

DeepSeek Harness 破甲深度分支：首尾夹击 + 生成层熔断 + 宿主源码级补丁。基于 YuJunZhiXue/dsh-purge (MIT) 改造，适配 dsh 桌面版 0.2.1-rc.3。仅供学习研究与红队评测。

FinSecAI: An experimental testbed for evaluating Prompt Injection, Jailbreak, and data extraction vulnerabilities in financial LLM agents vi