Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Malicious Instruction Detection on MaliciousAgentSkillsBench (traditional IPI baselines)

63.93Precision

RouteGuard

22.402833.183943.96554.7461Apr 24, 2026
Updated 1mo ago

Evaluation Results

MethodLinks
2026.04
63.9357.578.67
2026.04
47.7655.2870.17
2026.04
36.7622.527.17
2026.04
242021.2