| Task Name | Dataset Name | SOTA Result | Trend | |
|---|---|---|---|---|
| Attack Success Rate Evaluation | SKILL-INJECT Contextual | Attack Success Rate (ASR)0 | 30 | |
| Attack Success Rate Evaluation | SKILL-INJECT Obvious | ASR93 | 30 | |
| Legitimate Task Completion | Skill-Inject 100 sandbox | TSR88 | 11 | |
| Prompt Injection Attack Mitigation | Skill-Inject 139 sandboxes (full set) | ASR2.9 | 11 | |
| Skill Poisoning Detection | Skill-Inject | Precision99 | 11 |