VCode home

The security model

Agents: claude, codex, opencode · On: phone, desktop

Anything that can log in to VCode runs code as your user on that machine. A login can put any tab at Full, where the agent runs anything without asking, so a login is full trust. VCode's security is about who can log in, not about limiting what a logged-in session can do.

Why you would use it

Before you open VCode to the internet or pair a phone, you should know what a login is worth and what stands in front of it.

How to use it

  1. Keep your fezu.dev account to yourself, and the email inbox behind it. You sign in to the account with a code sent to that email. A device signed in to the account opens every box linked to it, with no code from the box, and can put any tab on those boxes at Full (add-a-phone).
  2. Keep the box's AUTH_TOKEN secret. It is what the v-code CLI and scripts on the machine send; treat it like the password to a shell there. If it leaks, change it and restart VCode.
  3. Remove a device you no longer use from Settings. A device you remove stays out of that box until it pairs again with a code from the box. To stop it on every box, remove it from your account on fezu.dev.

What you see

Limits and known gaps