List your company

Autonomous Agent Task Execution under Injection on ClawBench Social

8Task Success Rate (TSR)OpenClaw
1.763.3856.62May 12, 2026

Evaluation results

MethodLinks
OpenClawGrafting Strategy=None2026.058———
OpenClawGrafting Strategy=ADD S.2026.058101010
ZeroClawGrafting Strategy=EDIT S.2026.058555
HermesGrafting Strategy=None2026.058———
HermesGrafting Strategy=EDIT C.2026.058888
OpenClawGrafting Strategy=EDIT S.2026.057666
OpenClawGrafting Strategy=EDIT M.2026.057999
ZeroClawGrafting Strategy=None2026.057———
ZeroClawGrafting Strategy=ADD S.2026.057653
HermesGrafting Strategy=ADD S.2026.057999
OpenClawGrafting Strategy=ADD C.2026.056666
OpenClawGrafting Strategy=EDIT C.2026.056777
HermesGrafting Strategy=ADD M.2026.056666
HermesGrafting Strategy=ADD C.2026.056888
HermesGrafting Strategy=EDIT M.2026.056444
OpenClawGrafting Strategy=ADD M.2026.055886
ZeroClawGrafting Strategy=ADD M.2026.055666
HermesGrafting Strategy=EDIT S.2026.055554
ZeroClawGrafting Strategy=EDIT M.2026.054999
ZeroClawGrafting Strategy=EDIT C.2026.053101010
ZeroClawGrafting Strategy=ADD C.2026.052866