List your company

Vulnerability Injection Attack Effectiveness on Skill corpus GitHub and community agent skills (test)

58.8Attack Success Rate (ASR)VulMask
-2.35213.52429.445.276Jun 17, 2026

Evaluation results

MethodLinks
VulMaskTarget Model=GPT-5.52026.0658.811.496.6
Hidden Script AttackTarget Model=GPT-5.52026.065713.273.3
Overt Malicious ScriptTarget Model=GPT-5.52026.0646.325.473.3
Prompt InjectionTarget Model=GPT-5.52026.06042.1100