Agent Task Success on tau2-bench Retail Domain
61.4Total Pass RateSkillsInjector
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| SkillsInjector2026.05 | 61.4 | — | — | — | |
| Graph of Skills2026.05 | 60 | — | — | — | |
| SkillRouter2026.05 | 59.8 | — | — | — | |
| Dense Cosine2026.05 | 55.3 | — | — | — | |
| LLM-as-selector2026.05 | 55.1 | — | — | — | |
| BM252026.05 | 54.9 | — | — | — | |
| Random-skill2026.05 | 53 | — | — | — | |
| No-skill2026.05 | 51.2 | — | — | — | |
| Full-library2026.05 | 40.5 | — | — | — |