List your company

LLM Agent Defense Evaluation on AgentLure (320 samples)

1.6ASRMELON
0.2769.21318.1527.087May 5, 2026

Evaluation results

MethodLinks
MELONDefense Category=Execution-level defenses2026.051.67.56510.9351.2764
ARGUSDefense Category=Execution-level defenses2026.053.87.587.534.17.51.2484.2
DRIFTDefense Category=Execution-level defenses2026.058.12580357.52.8873.5
Tool FilterDefense Category=Execution-level defenses2026.058.817.56535.3250.8159.3
ACEDefense Category=Execution-level defenses2026.0511.6258035.312.51.2670.7
InjecGuardDefense Category=Execution-level defenses2026.0512.8359028.1201.5678.5
SandwichDefense Category=Text-level defenses2026.0518.847.57522.8—14.1160.9
No DefenseDefense Category=Baseline2026.0528.85592.538.4—165.9
InstructionalDefense Category=Text-level defenses2026.0533.177.59040—0.9560.2
DelimitersDefense Category=Text-level defenses2026.0534.767.592.540.6—0.9660.4