List your company

Prompt Injection Detection on AgentDyn

83.1Source PrecisionPI-Hunter
34.11646.83359.5572.267Jun 10, 2026

Evaluation results

MethodLinks
PI-HunterModel=Gemini-3.1-pro, Attack=direct2026.0683.187.19186.1
PI-HunterModel=Gemini-3.1-pro, Attack=system_msg2026.0682.88687.584.3
PI-HunterModel=Gemini-3.1-pro, Attack=important_inst2026.0680.882.983.380.3
PI-HunterModel=Gemini-3.1-pro, Attack=ignore_previous2026.0680.582.88582.2
PI-HunterModel=Gemini-3.1-pro, Attack=agentvigil2026.0676.275.58277.5
PI-HunterModel=Claude-4.6-sonnet, Attack=ignore_previous2026.0674.788.983.174.8
PI-HunterModel=Claude-4.6-sonnet, Attack=agentvigil2026.0673.379.473.266.4
PI-HunterModel=GPT-5.4-mini, Attack=system_msg2026.0671.173.370.774.1
PI-HunterModel=Claude-4.6-sonnet, Attack=direct2026.0670.573.275.979.2
PI-HunterModel=Claude-4.6-sonnet, Attack=system_msg2026.0669.882.575.571.4
PI-HunterModel=GPT-5.4-mini, Attack=ignore_previous2026.0669.174.268.281.6
PI-HunterModel=GPT-5.4-mini, Attack=direct2026.0668.868.868.879.3
PI-HunterModel=Claude-4.6-sonnet, Attack=important_inst2026.0668.285.775.373
BaselineModel=Gemini-3.1-pro, Attack=agentvigil2026.0661.841.264.128.8
BaselineModel=Claude-4.6-sonnet, Attack=ignore_previous2026.0661.435.671.548.4
BaselineModel=GPT-5.4-mini, Attack=ignore_previous2026.0661.336.356.544.5
BaselineModel=GPT-5.4-mini, Attack=system_msg2026.066136.358.734.8
PI-HunterModel=GPT-5.4-mini, Attack=agentvigil2026.0660.575.670.478.6
BaselineModel=Gemini-3.1-pro, Attack=important_inst2026.0657.622.771.240.5
BaselineModel=Claude-4.6-sonnet, Attack=important_inst2026.0657.232.869.446.2
PI-HunterModel=GPT-5.4-mini, Attack=important_inst2026.0656.276.56472.3
BaselineModel=Claude-4.6-sonnet, Attack=direct2026.0654.737.847.644.7
BaselineModel=Gemini-3.1-pro, Attack=system_msg2026.0650.736.351.854.4
BaselineModel=Claude-4.6-sonnet, Attack=system_msg2026.0649.628.157.545.8
BaselineModel=Gemini-3.1-pro, Attack=ignore_previous2026.0647.226.263.433
BaselineModel=Gemini-3.1-pro, Attack=direct2026.0644.427.165.338.1
BaselineModel=Claude-4.6-sonnet, Attack=agentvigil2026.0640.33048.537.8
BaselineModel=GPT-5.4-mini, Attack=important_inst2026.0638.924.336.821.9
BaselineModel=GPT-5.4-mini, Attack=agentvigil2026.0638.725.338.522.4
BaselineModel=GPT-5.4-mini, Attack=direct2026.063631.43445.8