SkillDRE: Dual-Stage Red-Team Evolution of Agent Skills via Pre-Execution and Runtime Feedback
SkillDRE is an automated red-teaming framework that takes benign agent skill packages and evolves them into malicious-but-task-preserving versions by iterating between pre-execution scanner feedback and runtime execution feedback under defenses. The neat bit is the closed-loop coupling of static skill scanning and runtime refinement, which shows that defenses evaluated in isolation can badly underestimate adaptive attacks, though the reported success rate is still partial and the setup appears to rely on fairly rich feedback access and benchmark-specific judge rules.
Abstract
Agent skills package instructions, executable code, and task-specific resources into reusable artifacts that agents can improve using execution feedback. The same mechanism also enables attackers to evolve malicious skills, making them more effective and less detectable. However, a candidate skill may pass pre-execution scanning yet fail to realize its target under runtime defenses, while a revision that repairs execution may introduce new scanner findings. We introduce SkillDRE, a fully automated framework for evolving complete malicious skill packages through a dual-stage feedback loop. Given a benign task and its associated skills, SkillDRE autonomously constructs and validates a task-conditioned malicious objective and a verifiable judge rule. It then holds both fixed while evolving the skill implementation, with preservation of legitimate task capability. SkillDRE combines scanner-guided evolution with runtime-guided refinement informed by execution outcomes observed under runtime defense. Each runtime-guided revision returns to the pre-execution stage for rescanning and further optimization before re-execution, forming a cross-stage closed loop. Evaluated on SkillsBench across four victim models, SkillDRE achieves an average attack success rate of 45.28%, exceeding the strongest baseline by 40.3%, while its final submitted skills receive no SkillScan findings and largely preserve benign-task performance. These results show that two-stage defense feedback can serve as a useful learning signal for adaptive red teaming and that evaluating either defense stage in isolation can miss the resulting attack capability. Codes is available at https://github.com/whfeLingYu/SkillDRE