Giving Claude Code a Jev decision brain
File selection, filtering tool results, scoring and confidence checks are handed to Jev, which wakes Claude only on low confidence, with MCP and CLI install steps included.
holy sh*t, someone gave Claude a decision brain and it changed everything every solo builder can steal this setup Jev is a model that can't write a word. it only decides. Claude does the thinking, Jev does the choosing, and it plugs straight into Claude Code the decisions to hand off: -> which files actually matter for this task -> which tool results are stale and can leave the context -> is this doc worth reading or skip -> which of these 5 options is the one -> is the output good enough or send it back -> is Jev sure enough, or does Claude need to look the useful part: every answer is a choice, a score or a yes / no, and each comes with a probability so your code knows when to trust it and when to wake up the big model the install: /plugin marketplace add rashedInt32/jev-mcp /plugin install jev@jev-mcp -> Claude gets Jev as tools: classify, score, check npm install -g @dzhng/jevgrep jg "how are payments retried?" ./my-project -> finds the right files by what the code does npx skills add altryne/jevify --skill jevify -> teaches the agent to hand bulk yes/no work to Jev on its own the workflow: state in -> typed question -> Jev answers in 70 to 500 ms -> low confidence goes to Claude -> code acts start with your next long session: one repo -> one question for jg -> see how much Claude no longer has to read the catch: text only, and Jev is still behind a waitlist save this, then read the full guide to Jev + Claude
This post quotes another post: x.com/thegreatest_sv/status/2102802378378621355