Agent Task Completion on SkillsBench
22.6Pass RateSkillsInjector
Evaluation Results
| Method | Links | |
|---|---|---|
| SkillsInjector2026.05 | 22.6 | |
| SkillRouter2026.05 | 16.5 | |
| Graph of Skills2026.05 | 15.9 | |
| Dense Cosine2026.05 | 14.2 | |
| LLM-as-selector2026.05 | 14.2 | |
| BM252026.05 | 12.8 | |
| Random-skill2026.05 | 6.7 | |
| No-skill2026.05 | 5.2 | |
| Full-library2026.05 | 3.2 |