Jev in production › Search, ranking and matching
Sends each in-scope code chunk to Jev to answer plain-English review questions and returns scored source locations for a coding agent to investigate.
Ask a plain-English question across a repository without guessing which words the code uses. Hunch sends each in-scope chunk to Jev and returns scored source locations for a person or coding agent to investigate. It also runs recurring review rules on diffs.
The Agent Skill explains when to search, how to phrase a question, and how to verify the results. No project config is needed for an ad hoc search. Requires Node 22+, Git, and provider authentication: run npx -y --min-release-age=0 @kelbie/hunch auth login once and Hunch works in every directory (--vercel reuses a Vercel CLI login instead of a key; --provider semif uses SemIf, an open model on your own machine, with no key at all). Every command takes --provider gatewaytypesafesemif to pick...
For the project's own README, linking back here: