Would jev help reduce workload from Claude?

[ed-hch]

Would jev helps reduce token use/claude by takeover the initial job matching/evaluate fit?

[MadsLorentzen]

Short answer: not in the core template. Jev is a separate model reached through a paid API, and two settled lines apply. Everything in this repository is free at the point of use, so nothing in the tree bills per call. And the workflow runs on Claude Code alone, with no second model backend. If token use in the first matching pass is the concern, /rank is already the cheap pass: it scores from the stored listing fields and only the shortlist gets the full evaluation. A fork is free to swap that step for anything it likes.