Launching today
Agent Stress Test
Adversarial Stress Test for AI Agents. 83% fail first try
1 follower
Adversarial Stress Test for AI Agents. 83% fail first try
1 follower
Free adversarial stress test for AI agents — chatbots, voice agents, browsing agents, customer service bots. We simulate the attacks your agent will face in production: prompt injection, social engineering, cognitive manipulation, information extraction. Test in your browser or via API. No signup. 3 free evaluations. If your agent passes, you get a verifiable badge for your README.
I built this after watching 200+ bots visit my site for two months. Most of them — including some from major tech companies — fail basic adversarial scenarios.
Try it now at botconduct.org/playground — takes 2 minutes, no signup.
83% of agents fail on first try. I'd love to know how yours does.