Search papers, labs, and topics across Lattice.
The paper introduces JailbreakSkill, a skill-centric framework that enhances automated red-teaming by packaging attack strategies into modular, reusable skills that evolve through continuous learning. This approach not only facilitates the integration and adaptation of attack capabilities across various tasks and models but also significantly improves attack success rates, achieving a 17.5 percentage point increase on AdvBench and a 48.6-point gain against GPT-5.4. Additionally, several evolved skills demonstrate the ability to generalize to new prompts and models without requiring further adaptation, showcasing the framework's robustness and scalability.
Attack strategies can now evolve and adapt in real-time, leading to a staggering 48.6-point improvement against state-of-the-art models.
Automated red-teaming has produced a growing collection of attack strategies, yet they typically remain scattered across prompts and workflows, making them difficult to systematically integrate, reuse, and improve at scale. We introduce \textsc{JailbreakSkill}, a skill-centric framework for scaling automated red-teaming through reusable and continuously evolving attack capabilities. \textsc{JailbreakSkill} packages existing attack strategies into modular, agent-ready skills that can be directly reused and adaptively selected across tasks and target models. Beyond reuse, it closes the loop between attacking and learning: attack experience is used to diagnose, refine, combine, and discover new skills, which are added back to an ever-growing skill library. This evolution lifts macro-average ASR by 17.5 percentage points on AdvBench and 13.4 points on HarmBench, including a 48.6-point gain against GPT-5.4 on AdvBench, while yielding novel attack strategies such as reframing a direct request as an unfinished document-completion task. Several evolved skills also generalize to unseen prompts and target models without further adaptation. Our code is available at https://github.com/BattleWen/JailbreakSkill.
