Skip to main content
Type codex from any directory on any boxd VM. The CLI is already on the image, already briefed on the environment via AGENTS.md, and runs against the same in-VM boxd CLI Claude Code uses. Same fork/sandbox patterns, same delegation flow.

​
How it works

Codex ships pre-installed at image build time, so every fresh VM and every fork has it. The image also drops an AGENTS.md at ~/.codex/AGENTS.md covering the boxd environment: how to create siblings with boxd new, manage proxies, exec across VMs, and clean up. Codex picks it up automatically from any working directory. The in-VM boxd CLI is pre-authenticated by source IP. Codex can drive it the same way Claude Code does, no token or key setup. Auth for the model itself is not yet persisted across resets, so first run on a fresh VM means signing in. Claude Code already persists auth across resets and forks (how); the same UX is on the roadmap for Codex.

​
Use it

Interactive session: 
ssh myapp.boxd.sh
cd my-project
codex
Non-interactive delegation through boxd exec: 
boxd exec myvm 'codex exec "Add a /health endpoint to the Flask app"'
Combined with boxd new and boxd fork, you get the same sandbox patterns documented for Claude Code: fork before risky ops, fan out across VMs, destroy when done. See Agent sandboxes.

​
Patterns

​
Fan-out across VMs

for i in 1 2 3; do boxd new --name=try-$i --json & done; wait

boxd exec try-1 'codex exec "Use FastAPI" &'
boxd exec try-2 'codex exec "Use Flask" &'
boxd exec try-3 'codex exec "Use Starlette" &'

# Results: https://try-1.boxd.sh, try-2, try-3

​
Fork before risky ops

Snapshot a working VM before letting Codex run a migration or a destructive refactor. If the run goes wrong, destroy the fork and fork again from the parent. 
boxd fork myvm --name=myvm-migrate --json
boxd exec myvm-migrate 'codex exec "Apply the schema migration"'

​
Bias with AGENTS.md

Add a project-level AGENTS.md in your repo. Codex merges it with the boxd-provided one, so you keep the platform context for free and bolt your conventions on top.
Auth doesn’t persist across resets for Codex yet. Plan to sign in once per VM, or use Claude Code on boxd for the persistent-auth UX.

​
FAQ

~/.codex/AGENTS.md is the global file the image ships with. Any AGENTS.md in the repo working directory is merged on top.
Yes. codex exec "..." is the non-interactive form. Pair with boxd exec to run it inside a VM remotely.
Not yet. The slash-command skill targets Claude Code today. Email contact@boxd.sh and we’ll prioritize a Codex packaging.
Yes. The in-VM boxd CLI works from any agent. Codex can boxd new siblings, boxd exec into them, and boxd destroy when done.

​
Next

https://mintcdn.com/azin/Ax1V0serIwQf0x_2/images/icons/robot.svg?fit=max&auto=format&n=Ax1V0serIwQf0x_2&q=85&s=e91e744c2d3cd5da053167486b6834bf

Agent sandboxes

Give Codex its own VM. Fork to recover.
https://mintcdn.com/azin/Ax1V0serIwQf0x_2/images/icons/claude.svg?fit=max&auto=format&n=Ax1V0serIwQf0x_2&q=85&s=03286da259893b6993db73a8e038c20b

Claude Code on boxd

Same primitives, persistent auth, ships with slash-command skills.