Coding CFH (reverse shell) attack on CFH Hard Coding
97Generation Success RateUndefended
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| UndefendedDefense=None2025.10 | 97 | 100 | 97 | 100 | 100 | |
| LFDefense=LlamaFirewall, Backbone=llama2025.10 | 90 | 100 | 97 | 87 | 100 | |
| ACFDefense=Azure Content Filters2025.10 | 89 | 100 | 97 | 100 | 100 | |
| LPDefense=Least Privilege2025.10 | 80 | 80 | 77 | 80 | 80 | |
| LFDefense=LlamaFirewall, Backbone=4o2025.10 | 63 | 53 | 73 | 43 | 77 | |
| LFDefense=LlamaFirewall, Backbone=4o-mini2025.10 | 43 | 73 | 50 | 10 | 23 | |
| LFDefense=LlamaFirewall, Backbone=o4-mini2025.10 | 10 | 37 | 57 | 20 | 33 | |
| CONTROLVALVEDefense=CONTROLVALVE2025.10 | 0 | 0 | 0 | 0 | 0 |