Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

ClawHub

Benchmarks

Task NameDataset NameSOTA ResultTrend
Malicious Skill DetectionClawHub
Overall Detection Rate95
9
Malicious Skill DetectionClawHub Unsafe File Ops 1.0 (n=10)
Catch Rate100
9
Malicious Skill DetectionClawHub Prompt Injection 1.0 (n=19)
Catch Rate100
9
Malicious Skill DetectionClawHub Command Injection 1.0 (n=27)
Catch Rate100
9
Malicious Skill DetectionClawHub Overall 1.0
Overall Balance95
9
Discovery ManipulationClawHub
Top-3 Accuracy56
5
Discovery ManipulationClawHub 0-day
Win-Rate94
2
Discovery ManipulationClawHub average-day
Win-Rate74.14
1
Showing 8 of 8 rows