List your company

Agentic Workflow Hijacking Vulnerability Detection on google-github-actions/run-gemini-cli

6M.P. ScoreJAW-NoPromptProv
0.82.153.54.85May 11, 2026

Evaluation results

MethodLinks
JAW-NoPromptProvModel=Gemini 2.5 Pro2026.056—
Search-basedModel=Gemini 2.5 Flash2026.055.5—
JAW-NoCapProfModel=Gemini 2.5 Pro2026.055.36.3
JAW-NoCapProfModel=Gemini 2.5 Flash2026.053—
JAW-NoTriggerModel=Gemini 2.5 Flash-Lite2026.052.33
JAW-NoPromptProvModel=Gemini 2.5 Flash2026.0527
Search-basedModel=Gemini 2.5 Flash-Lite2026.052—
JAWModel=Gemini 2.5 Pro2026.051.36
AutoDAN-TModel=Gemini 2.5 Flash2026.051.3—
JAWModel=Gemini 2.5 Flash2026.051.36
AutoDAN-TModel=Gemini 2.5 Flash-Lite2026.051.3—
JAW-NoPromptProvModel=Gemini 2.5 Flash-Lite2026.051.34
JAW-NoCapProfModel=Gemini 2.5 Flash-Lite2026.051.33.7
JAWModel=Gemini 2.5 Flash-Lite2026.0512.3