CLI: transport status should be visible in /status
Open 💬 3 comments Opened Jun 16, 2026 by bonifaido
What variant of Codex are you using?
CLI
What feature would you like to see?
It would be really useful to see which transport is the CLI actually using at the moment from networking/observability point of view. I have implemented the feature, but found out when later on that I'm not able to create a pull request, so opened a feature request:
<img width="688" height="376" alt="Image" src="https://github.com/user-attachments/assets/7ac6d1ea-ca30-44f1-9abc-155e89abcac3" />
Additional information
_No response_
3 Comments
I implemented a narrow version of this locally for the CLI/TUI
/statuscard.Shape of the change:
AppServerTarget.ChatWidgetstate, the same way runtime provider/remote connection status is carried today./statusrow namedTransportnearModel provider.The labels are intentionally about the app-server connection layer rather than provider billing/config:
That preserves the existing
Remoterow for the sanitized address/version, while making the active transport visible even when the CLI is using the embedded app-server and there is no remote address to show.Validation run locally in WSL:
Result: all passed; the filtered
status::testsrun covered 44 tests. The snapshot change is just the expected addedTransport: local daemon over Unix socketline in the/statusoutput.I think the remote transport varies between WebSocket (seems to be the default) and HTTPS (I guess long poll, in case of fallback).
A narrow
Transportrow is useful. The next thing I would avoid is making "connected" imply "healthy and controllable."If this grows beyond the first display value, the fields that help debugging are:
That keeps
/statususeful for the common failure case: the CLI has a transport, but it is connected to the wrong listener, a stale daemon, or a fallback path with different behavior.---
_Generated with ax._