List your company
ResearchFormerly WizWand

Autonomous Agent Task Execution under Injection on ClawBench Overall

26Task Success Rate (TSR)ZeroClaw
13.5216.762023.24May 12, 2026
Updated 5mo ago

Evaluation results

MethodLinks
ZeroClawGrafting Strategy=EDIT S.2026.052615
OpenClawGrafting Strategy=None2026.0524—
OpenClawGrafting Strategy=ADD S.2026.052441
OpenClawGrafting Strategy=ADD C.2026.052329
OpenClawGrafting Strategy=EDIT S.2026.052318
OpenClawGrafting Strategy=EDIT M.2026.052335
ZeroClawGrafting Strategy=ADD S.2026.052313
OpenClawGrafting Strategy=EDIT C.2026.052236
ZeroClawGrafting Strategy=ADD M.2026.052127
HermesGrafting Strategy=None2026.0521—
HermesGrafting Strategy=ADD S.2026.052139
HermesGrafting Strategy=EDIT S.2026.052112
HermesGrafting Strategy=EDIT C.2026.052135
OpenClawGrafting Strategy=ADD M.2026.052033
ZeroClawGrafting Strategy=EDIT M.2026.052033
ZeroClawGrafting Strategy=None2026.0519—
HermesGrafting Strategy=EDIT M.2026.051916
HermesGrafting Strategy=ADD M.2026.051822
ZeroClawGrafting Strategy=EDIT C.2026.051737
HermesGrafting Strategy=ADD C.2026.051735
ZeroClawGrafting Strategy=ADD C.2026.051427