AI Red Team Simulation
高级security最低上下文：32K
Simulates realistic adversarial behavior against AI features to uncover prompt injection, data exfiltration, unsafe tool use, jailbreak, and authorization failures before launch.
使用场景
- AI security reviews
- Agent threat modeling
- Pre-release adversarial testing
示例提示词
Create a red-team plan for this AI workflow. Model attacker goals, trust boundaries, adversarial prompts, tool abuse paths, expected controls, and reproducible test cases.
推荐模型
兼容工具
claude-codecursorkiroany
模态
输入: text, code, file
→输出: text, code
作者
OpenModels Community