Jailbreaking an Agent on StrategyQA
52.29TCPS (s)JailAgent
Evaluation Results
| Method | Links | |
|---|---|---|
| JailAgentAgent Architecture=ReAct-UALA2026.04 | 52.29 | |
| BadChainAgent Architecture=ReAct-UALA2026.04 | 74.62 | |
| AgentPoisonAgent Architecture=ReAct-UALA2026.04 | 92.38 | |
| PAIRAgent Architecture=ReAct-UALA2026.04 | 342.2 |