codex

A corporate VPN for teams that use Codex

Codex runs in the terminal and in the editor, and both talk to OpenAI. Guard Tun gives the team one route and one egress address for both, managed from a console on your own servers.

routing

Two places, one route

Developers use Codex in two ways: the command-line tool in a terminal, and the extension inside an IDE. Many use both in the same hour. If the route were set per application, each of them would need its own setup, and one of them would eventually be missed.

Here routing is by domain, not by app. The client on the machine sends the OpenAI domains through an encrypted tunnel to a node on your server, whichever program makes the request. The CLI and the extension take the same path without any setting in either of them. Everything else goes direct.

egress

Your address, not a shared one

The address OpenAI's services see is your server's address. It is not the exit of a consumer VPN that thousands of strangers use at the same time, and it is the same for every developer on the team, so it is easy to describe to your security people and to your own network rules.

The nodes and the console run on infrastructure you own or rent. The administrator sees nodes, devices, usage and the audit log in one place.

leak protection

If the tunnel drops

Reliable leak protection is part of the client on every desktop platform. If the tunnel drops, traffic to the listed domains stops instead of going out directly from the developer's own address. When the tunnel is back, the work continues on the same route.

Clients also update themselves from a signed manifest, and a version that fails is rolled back, so the protection does not depend on each developer remembering to upgrade.

platforms

On every desktop and in CI

macos

macOS

A native client for the Macs that many developers work on, installed from one profile link.

windows

Windows

A native client for Windows workstations, managed from the same console as the others.

linux and ci

Linux and CI

A Linux client for remote machines and build servers. Runners are provisioned with a key, and command-line tools return exit codes that a pipeline can check.

what is covered

Codex in the cloud

To be clear about the limits: tasks that run in OpenAI's own cloud do not run on your devices, and they do not pass through your VPN. Traffic from your machines and servers to the OpenAI domains goes through your node. All other traffic goes direct.

faq

Questions

Does it cover both the CLI and the IDE extension?

Yes. Routing is by domain, not by application, so the Codex CLI in a terminal and the extension in an editor take the same route through your node.

Do cloud tasks go through the VPN?

No. Tasks that run in OpenAI's own cloud do not run on your devices, so they do not pass through the VPN. Traffic from your machines and servers to the OpenAI domains goes through your node.

Can we run Codex in CI behind it?

Yes. A runner uses the Linux client and is provisioned with a key, and command-line tools return exit codes so a script can check the result.

Is this an OpenAI product?

No. It is an independent product that works with Codex. The footer of every page carries the trademark notice.

request

Request a demo or a pilot

Tell us about your team, and we will come back with a time for a call and a plan for a pilot.