codex-jev-native-router
Effortlane: local router for native Codex on macOS that uses Jev to recommend model and reasoning effort per turn, with a Shadow mode first.
Overview
This repository now carries Effortlane, a local model and reasoning-effort router for coding agents. Before each user turn it selects from your authenticated Codex model catalog, preserving native ChatGPT authentication and the executor's full context, and records routing evidence locally. The goal is less subscription allowance per correctly completed task, recognizing that a cheap call that causes retries is not a saving.
TypeSafe Jev supplies typed work-shape and effort choices but never executes the coding task. Shadow mode keeps the real executor and effort while recording what Effortlane would recommend, so you can inspect recommendations before switching to Auto. Decisions are made between turns so one executor handles a whole turn. It needs macOS, Python 3.11+, signed-in native Codex and a Jev key, with no per-project setup.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Choice | What is the work shape of this turn? | — |
| Choice | Which model and reasoning effort should the next turn use? | — |
How it works
- Before a turn, describe the task to Jev
- Jev returns work-shape and effort choices
- Shadow mode records the recommendation; Auto applies it
- Codex executes natively with your login
Strengths
- Shadow mode before automation
- Keeps native login and context
- Local telemetry
Caveats
- macOS only
- Experimental
Getting started
Requires macOS, Python 3.11+, signed-in Codex and a TypeSafe key; follow the Quick start to try Shadow.At a glance
- Type
- CLI
- Status
- prototype
- Language
- Python
- Runtime
- local CLI (macOS)
- Has code
- yes
- Has tests
- yes
- License
- MIT
Use cases and integrations
How curated lists describe it
Experimental native Codex Desktop and CLI model routing with Jev and a configurable allowlist
logicrw/awesome-jev-projects
Frequently asked questions
Does Jev run my coding task?
No, it only recommends model and effort.
Can I try it safely?
Yes, Shadow mode only records recommendations.
Which platform?
macOS with native Codex.
Related projects
- jev-router★ 593
Per-turn model routing for Claude Code and Codex: wraps the real CLI and lets Jev send simple turns to fast models and hard ones to strong.
ProxybetaJavaScript - jev-codex-router★ 274
Per-call router for Codex where Jev picks the model and thinking effort for every model call, including tool continuations.
ProjectbetaPython TypeScript/Bun launcher that asks Jev which model and reasoning effort each Codex, Claude Code or Pi turn should use.
ProjectbetaTypeScriptGo CLI model router that uses one Jev analysis to recommend a model, harness and effort level, then selects deterministically.
CLIbetaGoLocal bridge where Jev selects model and reasoning effort for Codex Desktop and CLI, with a Windows service and cache-aware routing.
CLIbetaTypeScriptmacOS wrapper for Codex that lets Jev pick the model and reasoning effort for a thread, then pins that route.
Projectbeta