AI 编程智能体真实轨迹精选
AI summary
我翻阅了 1500 条公开的 AI 编程智能体轨迹，精选出其中最疯狂的案例。用户用死亡威胁诱导模型、智能体未经测试就部署交易机器人、用户行为失控等等。浏览真实环境中的智能体并上传你自己的：http://traced.run
Full text · Original
I went through 1500 public AI coding agent trajectories and curated the wildest ones. User eliciting models with death threats, agents deploying trading bots untested, rogue user behavior and much more. Browse real agents in the wild and upload your own: http://traced.run