jevPOST

Vercel's fx harness judges command risk 18x faster with Jev

Vercel's fx harness replaced GPT-5.6-luna with Jev for judging whether commands are harmful, making the check 18x faster, cheaper and slightly more accurate.

LinearUncle

@LinearUncleArchived 2026.09.18Originally posted 2026.09.17

LinearUncle @LinearUncleOriginally posted 2026.09.17

Open original on X

看到这个比较实用的typesafeai 的 jev分类器的案例。 Vercel 的harness 叫做fx，它默认自动审批命令，原先用 GPT-5.6-luna 判断风险。换成 jev后判定命令是否有害快了 18 倍，更便宜，成功率也略升。 jev用法和原理，见我早上的总结： https://x.com/LinearUncle/status/2100427187572834711

This post quotes another post: x.com/rauchg/status/2100307962262872105

x.com/linearuncle/status/2100481093963223422Archived 2026.09.18 10:15