foreman-jev
Experimental Foreman derivative where Jev supervises a Codex worker via Vercel AI Gateway, with programmer-chosen acceptance commands.
Overview
foreman-jev is an experimental derivative of thruwire/foreman from Shifty Eye Games. Jev, reached through Vercel AI Gateway, supervises a Codex coding worker, while a programmer-selected acceptance command checks the result. The public release preserves a runtime that passed a fresh macOS install and a real Codex and Jev coding run, with documented evidence and platform limits.
It also includes a read-only Azure DevOps PR review workflow: it collects an immutable diff, accepts a review draft from a programmer or ChatGPT, and asks Jev to assess the draft's evidence, producing local reports without launching a reviewer, approving or merging. Tests and a demo run offline; real runs need an authenticated Codex CLI and an AI Gateway key with Jev access. The pilot targets macOS and Linux, with Windows users directed to WSL2.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Yes / no | Is the Codex worker's task complete and verified?(inferred) | yes · no |
| Yes / no | Does the review draft's evidence support its findings?(inferred) | yes · no |
How it works
- foreman run with a repo, job and --verify command
- Codex implements; Jev supervises progress and completion
- Acceptance command checks the result
- Optional read-only PR review assessment
Strengths
- Offline tests and demo
- Programmer-controlled acceptance command
- Read-only PR review mode
Caveats
- Experimental pilot
- Needs Codex CLI and Vercel AI Gateway credits
Getting started
git clone, python -m pip install '.[dev]', python -m pytest -q, foreman demo --repo .; then foreman run --repo ... --job ... --verify '...'.At a glance
- Type
- Project
- Status
- prototype
- Language
- Python
- Runtime
- CLI
- Has code
- yes
- Has tests
- yes
- License
- MIT
Use cases and integrations
How curated lists describe it
An experimental Jev supervisor for Codex workers with programmer-selected acceptance commands.
logicrw/awesome-jev-projects
Frequently asked questions
How does it relate to Foreman?
It is an experimental derivative of thruwire/foreman.
How is Jev accessed?
Through Vercel AI Gateway.
Does the PR review approve anything?
No, it only produces local reports.
Related projects
Claude Code plugin with rules and checks that make agents prove their work; an optional Jev judge triages review findings.
PluginbetaMarkdown + shellSoftware-factory supervisor where Jev independently judges coding agents' completion, test sufficiency, progress and need for a human.
ProjectbetaPythonClaude Code skills that wait for AI pull-request reviewers and judge whether their reports actually permit a merge, never inferring a pass.
ProjectbetaDesktop app running coding agents in parallel git worktrees, with optional Jev judgments of each changed file against the task.
AppbetaStop hook for Claude Code and Codex that refuses 'done' until a check passes after the last edit; Jev advises but never causes a block.
CLIbetaTypeScriptClaude Code plugin that scores a turn's redacted diff on eight Jev risk axes at Stop and asks Claude to reinvestigate once if high.
ProjectbetaGo