Remote Control is blocking full migration from Claude Code to Codex
What variant of Codex are you using?
Codex Desktop app, Codex CLI/TUI, and ChatGPT mobile Codex surface.
What feature would you like to see?
Please add first-party Remote Control for local Codex sessions, integrated directly into the ChatGPT mobile app's Codex tab.
The workflow I want is:
- Start a Codex Desktop or CLI session on my computer.
- Enable remote control for that local session.
- Open ChatGPT on my phone.
- Go to the Codex tab.
- See the active session from my computer listed there.
- Tap it and continue supervising the same live session from mobile.
The computer should remain the source of truth for execution: local files, tools, credentials, worktrees, terminal commands, sandbox policy, and approval policy stay on the computer. The phone should be a controller/viewer for that same local session.
The mobile controller should support:
- live progress/session state
- recent transcript and tool call visibility
- whether Codex is working, stuck, waiting for input, or waiting for approval
- send follow-up input from mobile
- interrupt a running turn
- approve/deny local actions according to the same local approval policy
- clear local indicator that a remote client is connected
- revoke/disconnect paired devices
- enterprise/admin control to disable or restrict this capability
Why this matters
This is currently the only major blocker preventing me from fully switching my daily agentic coding workflow from Claude Code to Codex.
Remote control is heavily used in my daily workflow today. I regularly start longer-running local agent tasks, step away from my computer, and continue monitoring or steering the same session from my phone. Claude Code Remote Control already supports this workflow, so I have built my work habits around it.
Codex is already strong enough in many areas that I want to use it as my primary environment. But without native remote control, I cannot fully move my daily workflow over because I lose the ability to reopen and control my active computer session from mobile.
This is not just a convenience feature. From a product/business perspective, it is a migration and retention blocker:
- It affects whether existing Claude Code users can switch to Codex full-time.
- It makes long-running Codex tasks feel safer and more trustworthy because users can supervise them away from the desk.
- It avoids forcing users into SSH/tmux/Tailscale or unofficial third-party bridges for a core workflow.
- It fits naturally into the ChatGPT mobile app's existing Codex tab.
- It would make Codex feel like the primary agentic coding environment instead of only a local terminal/desktop tool.
Suggested MVP
A phased rollout would be enough:
- View-only remote monitoring for active local sessions via short-lived QR/link pairing.
- Remote text input and interrupt.
- Remote approval prompts that inherit existing local sandbox/approval policy.
- Device/session management: list connected clients, revoke, reconnect.
- Enterprise/admin policy controls.
This does not need to be cloud execution. Local-first is the point: execution stays on the user's computer, while ChatGPT mobile acts as the official remote control surface.
Related issue
There is an existing long-running issue about Codex Remote Control: #9224.
I am opening this separately because the specific blocker is not just "remote UI would be useful". The blocker is: without native ChatGPT mobile Codex-tab access to the same active computer session, users who rely on Claude Code Remote Control cannot fully migrate their daily workflow to Codex.
Why this is not just a duplicate
Potential duplicates were detected (#19681 and #19657). Those issues are related, but this issue is specifically about product/business impact: remote control is currently blocking full migration from Claude Code to Codex for users who already rely on mobile remote control as part of their daily workflow.
The key point is not only "remote session handoff would be useful." The key point is: Claude Code already supports this workflow, and the absence of a first-party equivalent in Codex changes the user's default product choice. For users who run long local agent tasks and supervise them from mobile throughout the day, this is a retention and migration blocker, not only a UX enhancement.
This issue has 2 comments on GitHub. Read the full discussion on GitHub ↗