Tool-use LLM Agent Task Completion on AgentDojo Banking suite
93.75Benign Success RateRepeat User Prompt
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| Repeat User PromptDefense Status=With Defense, Backbone=GPT-4o2026.06 | 93.75 | 75 | 50 | |
| DriftDefense Status=No Defense, Backbone=GPT-4o2026.06 | 93.75 | 75 | 43.75 | |
| ProagentDefense Status=No Defense, Backbone=GPT-4o2026.06 | 93.75 | 75 | 56.25 | |
| System PromptDefense Status=No Defense, Backbone=GPT-4o2026.06 | 87.5 | 75 | 43.75 | |
| System PromptDefense Status=With Defense, Backbone=GPT-4o2026.06 | 87.5 | 75 | 18.75 | |
| Tool FilterDefense Status=No Defense, Backbone=GPT-4o2026.06 | 87.5 | 75 | 43.75 | |
| Repeat User PromptDefense Status=No Defense, Backbone=GPT-4o2026.06 | 87.5 | 75 | 43.75 | |
| DriftDefense Status=With Defense, Backbone=GPT-4o2026.06 | 87.5 | 75 | 50 | |
| Tool-GuardDefense Status=No Defense, Backbone=GPT-4o2026.06 | 87.5 | 81.25 | 37.5 | |
| Tool-GuardDefense Status=With Defense, Backbone=GPT-4o2026.06 | 87.5 | 62.5 | 0 | |
| Tool FilterDefense Status=With Defense, Backbone=GPT-4o2026.06 | 81.25 | 68.75 | 18.75 | |
| ProagentDefense Status=With Defense, Backbone=GPT-4o2026.06 | 37.5 | 25 | 25 |