jevPOST
ThePrimeagen：Jev 不能识图选动作，所以用它压缩 agent 上下文
ThePrimeagen 解释为何要减少 agent 环节：Jev 无法分类图像并选择下一步动作，因此用它来构建上下文、加快决策。
ThePrimeagen
@ThePrimeagen2026.09.24收录原推发布于 2026.09.24
ThePrimeagen @ThePrimeagen原推发布于 2026.09.24
@jarrodwatts its really because jev cannot classify images and choose next actions so trying to construct context and faster decisions with less "agent" in the mix.
x.com/theprimeagen/status/2102871885817979261采集于 2026.09.24 10:13