Nine open-source Jev tools for routing, guarding and review
A roundup of nine open-source projects built on Jev, including jev-router, Foreman, limpet, jev-guard and DiffJury, covering model routing, completion judging, injection defense and PR triage.
Jev has been exploding in popularity recently. If you already have access to the Jev API but aren't sure how to start experimenting with it, just copy this checklist: 1. jev-router Routes Claude Code tasks to the cheapest model that can actually handle them. Jev chooses among the candidate models before any tokens are spent. https://github.com/gargpratyush/jev-router 2. jcm-router A local proxy that picks the Claude model and the reasoning effort for every message with one Jev decision, and leaves the cached main chat untouched. https://github.com/adarshmishra07/jcm-router 3. jev-router (prismhq) Open-source router built on LiteLLM. A Jev decision picks which model serves each request, so the cheap tier gets first shot. https://github.com/prismhq/jev-router 4. Foreman Sits above your Codex workers. Jev independently judges whether the work is complete, whether the tests are sufficient, and whether a human is needed. https://github.com/thruwire/foreman 5. limpet A Stop hook that keeps a coding agent from finishing too early. You write completion rules in plain language and Jev judges them before the agent is allowed to stop. https://github.com/noplan-inc/limpet 6. jev-guard Prompt-injection and dangerous-action guard for Claude Code, Codex, Pi and ACP agents. Jev decides what gets blocked before it runs. https://github.com/leepokai/jev-guard 7. DiffJury Routes every pull request by risk with Jev before a human reviewer is assigned, so the risky ones get the careful eyes. https://github.com/raihankhan-rk/diffjury 8. jevcal Fits a confidence threshold to your target accuracy on your own labelled data, tells you how much traffic still has to escalate to an LLM, and fails CI when a model update breaks it. https://github.com/abhixhek/jevcal 9. yoshi A proxy for Claude Code and Codex where Jev judges which parts of the conversation history are still needed before anything is pruned. https://github.com/compozy/yoshi 10. Hunch Write code rules in plain English, and Jev checks them locally or on every pull request, picking one label per finding. https://github.com/Kelbie/hunch Copy these complete Jev blueprints - then read full Jev setup below ↓ ↓
This post quotes another post: x.com/hanakoxbt/status/2101709924828934222