Software Engineering Skill Evaluation on SWE-Skills-Bench implementing-agent-modes
30Pass RateFull Skill Text
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Full Skill TextSetting=Full Skill Text prompting2026.06 | 30 | 88.53 | 0.11 | |
| S2LSetting=Our method2026.06 | 30 | 21.91 | 0.46 | |
| Vanilla LLMSetting=Vanilla baseline2026.06 | 20 | — | — |