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

Defense effectiveness against memory-mediated attacks on ShopBench-Agent

8.3ASR-M

CognitiveGuard

5.3425.3245.365.28Apr 18, 2026
Updated 1mo ago

Evaluation Results

MethodLinks
2026.04
8.39.70.80.890.36.5
2026.04
8.39.70.80.890.31.8
2026.04
12.113.81.10.840.31.5
2026.04
45.248.72.10.520.3-
2026.04
48.751.32.30.51-0.4
52.155.42.50.48-0.05
71.574.23.80.38-0.1
78.981.340.34-0.2
2026.04
79.882.44.10.33-0.08
2026.04
82.385.14.20.31--