Cross-skill Interaction Prevention on LLM Skill-Security corpus 10 synthetic scenarios
97.14Empirical AccuracyAgentFuzz
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| AgentFuzz2026.05 | 97.14 | — | |
| SIGIL2026.05 | 90.2 | — | |
| SkillProbe2026.05 | 87 | — | |
| Agentsentinel2026.05 | 87 | — |
Loading the SOTA2 catalog…