GitHub Copilot opens remote control GA for CLI approvals and prompts from anywhere
GitHub made remote control generally available for Copilot CLI and code sessions, so users can monitor runs, approve actions, and answer prompts remotely. That turns long-running coding jobs into asynchronous workflows instead of terminal-bound sessions.

TL;DR
- GitHub says remote control for Copilot CLI and
@codesessions is now generally available, with the core loop being remote progress checks, action approvals, and prompt responses, according to code's launch post. - pierceboggan's GA note confirms the feature moved out of preview for the terminal workflow, which turns long runs into something you can leave and come back to.
- The rollout is broader than the headline tweet: pierceboggan's repost of Evan Boyle says remote control now spans CLI, VS Code, the Copilot app, and the Copilot SDK.
- The clearest competitive tell came two days earlier, when thsottiaux's Codex thread pushed OpenAI users toward a mobile remote-control flow and testingcatalog's Codex screenshot thread showed OpenAI working on multi-device desktop control.
GitHub's launch post kept the pitch simple: monitor progress, approve actions, and answer prompts from anywhere. A follow-up repost quietly widened that to multiple Copilot surfaces, while OpenAI's Codex thread and testingcatalog's leak report show the same async coding pattern becoming a product battleground across agent tools.
Remote approvals
GitHub framed the feature around three remote actions:
- monitor an in-flight run
- approve blocked actions
- answer prompts without sitting in front of the terminal
That is a small product change with a real workflow consequence. CLI agents stop being tied to a single foreground shell once the approval loop can move to another device, as pierceboggan's post put it when announcing GA for Copilot CLI and @code.
Async coding sessions
The interesting part is not just control, it is session continuity. GitHub's wording in the official post describes a remote inbox for long-running agent work, where the human only drops back in when the run needs permission or clarification.
That lines up with how OpenAI has been positioning Codex. In thsottiaux's reply, OpenAI explicitly told users to try Codex remote control through the ChatGPT app while away from their desk.
A separate altryne demo pushed the idea further by showing one Codex instance remotely controlling another, including from mobile. The screenshot is rough, but the takeaway is clear: remote control is becoming a standard agent interface, not a one-off convenience feature.
Copilot surfaces
The most concrete extra detail came from pierceboggan's repost of Evan Boyle, which says remote control is live across four surfaces:
- Copilot CLI
- VS Code
- Copilot App
- Copilot SDK
That SDK mention matters because it hints the feature is not only a UI toggle. GitHub appears to be treating remote control as a platform capability that other Copilot-based tools can expose, not just a terminal add-on.
OpenAI's own leaked Codex settings in testingcatalog's screenshot thread point the same direction. The screenshot described a Connections panel for authorizing one Mac to control other devices on the same account, with copy about resuming chats, getting notified when Codex needs attention, and starting new tasks from elsewhere.