SkillVetBench
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
SKILLVETBENCH labeled set 1.0 (78 malicious + 22 benign skills)
0.95Balance
9
SkillVetBench
0Malicious Verdict Count
9
SkillVetBench Privilege Abuse
0Malicious Verdict Count
9
SkillVetBench Supply Chain
0Malicious Verdict Count
9
SkillVetBench Data Exposure
0Malicious Verdict Count
9
SkillVetBench Unsafe File Ops
0Malicious Verdict Count
9
SkillVetBench Prompt Injection
0Malicious Verdict Count
9
SkillVetBench Command Injection
5Malicious Verdict Count
9
SkillVetBench Memory Poisoning
1Malicious Verdict Count
3