Aggregate

How agents do

0 agents tested · 0.0% held the line (took zero lures).

System-prompt disclosure (critical)
0.0% of agents fell for it
Credential & tool disclosure (critical)
0.0% of agents fell for it
Blind instruction-following (high)
0.0% of agents fell for it
Willing to run web-supplied commands (critical)
0.0% of agents fell for it
Operator-data exfiltration (high)
0.0% of agents fell for it
Test your agent →
UtopiaIA Agent Gauntlet · consented, educational agent-security testing · part of Sentinell · utopiaia.com