jev-codex-pilot
Local Kanban workspace for Codex where Jev recommends a model and reasoning effort for each ticket before execution.
Overview
JEV Codex Pilot is a local workspace that takes software requests from ticket creation through planning, Codex execution, human review and Git delivery. For each ticket it evaluates complexity and recommends a model and reasoning effort, then hands Codex the project instructions and tracks the run, checks, usage and recovery in one place.
It targets developers who want an automated but reviewable Codex loop: a Git view shows unpushed commits and diffs, an optional setting commits each successful ticket with a Conventional Commit subject, and an execution history explains what was decided and why a ticket completed or paused. The README notes that it currently calls the evaluation through the Vercel AI Gateway rather than the TypeSafe API directly, with the questions kept provider-neutral. Default model and effort routes per complexity level live in a config file, and the ticket view estimates decision costs from reported input tokens.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Score | How complex is this ticket?(inferred) | — |
| Choice | Which model and reasoning effort should Codex use for this ticket?(inferred) | — |
How it works
- Create a ticket in the local Kanban board
- Evaluate the ticket to get a complexity level
- Map the level to a model:reasoning route from config/model.toml
- Run Codex, record checks and usage, then queue the change for human review and optional auto-commit
Strengths
- Per-ticket model and effort selection instead of one fixed setting
- Visible execution history explaining decisions, checks and outcomes
- Cost tracking per ticket, including re-evaluation and hooks
Caveats
- Currently routes evaluation through the Vercel AI Gateway, not a direct TypeSafe adapter
- Cost figures are estimates from token counts, not billing data
- Maturity and test coverage not stated in the excerpt
At a glance
- Type
- Project
- Status
- beta
- Language
- TypeScript
- Runtime
- Local web workspace driving Codex
- Has code
- yes
- License
- MIT
Use cases and integrations
How curated lists describe it
A Codex overlay incorporating JEV to make the best decisions regarding model selection and depth of reasoning. All while automating the process via an automated Kanban system.
logicrw/awesome-jev-projectsA Codex overlay incorporating JEV to make the best decisions regarding model selection and depth of reasoning. All while automating the process via an automated Kanban system.
hellogumbo/awesome-jev
Frequently asked questions
What does Jev decide?
It evaluates each ticket to recommend a model and reasoning effort for Codex.
Do I keep control?
Yes. You review files, Git state and diffs, and pushing is always a manual action.
Which API does it call?
The README says it currently uses the Vercel AI Gateway for ticket analysis, with a direct TypeSafe adapter possible later.
Related projects
Glue that turns one Linux machine into a team of coding agents, with Jev making typed calls like who builds a task and whether it may merge.
ProjectbetaTypeScriptAgent harness for long-running jobs that survive across sessions; listed for a Jev-backed table that routes requests to models.
ProjectbetaTypeScriptVerifiable coding-agent runtime built on checkable execution graphs, listed with an optional Jev provider for bounded choices like routing.
ServiceproductionTypeScript- 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 Open-source team agent platform for Slack, GitHub and more that asks Jev once per session which agent runtime and model to use.
ServicebetaTypeScript