Your GitHub Copilot Workflows, Now From Any Device: A Remote Control Guide

By • min read

Overview

Modern development rarely follows a linear, single-task path. A typical session might involve an AI agent refactoring code in your editor, another debugging tests in the terminal, and a third scaffolding a new feature — all simultaneously. For a long time, managing this multi-agent workflow meant staying tethered to your desk. The moment you walked away from your laptop, you lost visibility into every running session.

That limitation has been removed. GitHub Copilot’s new remote control capability allows you to take your coding sessions anywhere — via github.com, the GitHub Mobile app, and directly within VS Code or JetBrains IDEs. This guide walks you through everything you need to set up and use remote control, from starting a session to completing a pull request from your phone.

Prerequisites

Before you begin, ensure you have the following:

• A GitHub account with an active GitHub Copilot subscription (Individual, Business, or Enterprise).
• GitHub Copilot CLI installed (if you plan to start sessions from the terminal). Alternatively, the latest version of the GitHub Copilot extension for VS Code (minimum version required for remote control is automatically updated).
• GitHub Mobile app (optional, for mobile monitoring and control).
• Any repository you want to work with — remote control works with both repositories and directories without repositories.

No additional installation is needed if you already have the latest Copilot CLI or VS Code extension. Simply ensure your GitHub account is authenticated on all devices you intend to use.

Step-by-Step Instructions

1. Starting a Copilot Session

Begin a standard Copilot session wherever you normally work. For example:

• In VS Code, open the command palette and run “Copilot: Start Agent Session”. Or use the integrated chat to begin an agent-based task.
• In Copilot CLI, run a command like copilot plan "refactor the auth module" or copilot explain to initiate a session. The session will appear in your current terminal.
• In JetBrains IDEs, use the Copilot tool window to start a session.

Once the session is active, you can proceed to make it accessible remotely.

2. Sending a Session to the Web or Mobile

To transfer your current session from a local environment to the cloud (so you can monitor and interact from other devices), use the /remote on command. This command works within any active Copilot session, regardless of the starting tool.

• In Copilot CLI: While a session is running, type /remote on and press Enter. The system will confirm and provide a URL pointing to the session on github.com.
• In VS Code or JetBrains: Open the Copilot chat panel, type /remote on and send it. A pop-up or message will appear with the link to the web view.

Once enabled, the session appears in real time on github.com under the “Copilot” section. You can also access it from the GitHub Mobile app by navigating to your Copilot sessions.

3. Monitoring in Real Time

Open the session URL on any other device — a phone, tablet, or another computer — to see exactly what the agent is doing. The live view includes:

• The plan the agent is researching or generating.
• Files it is reading or modifying.
• Diffs (changes) it is making to the code.
• Commands it is executing in the environment.

All updates stream in nearly real time, giving you full transparency even when away from your primary machine.

4. Steering a Session Mid-Flight

Remote control isn’t just for watching; you can interact. If the agent starts moving in a wrong direction, or you want to expand the scope, send a follow-up instruction directly from the remote interface:

• On github.com, use the chat input at the bottom of the session view.
• On the GitHub Mobile app, tap the message input to type natural language instructions.

Common steering commands include:

• “Please also update the test cases for this module.”
• “Stop, that approach is too complex. Use a simpler pattern.”
• “Can you create a pull request once you finish?”

Additionally, you can approve or deny permission requests (e.g., file writes, command execution) from the remote view, maintaining security while staying mobile.

5. Completing the Full Workflow from Anywhere

Remote control enables an end-to-end development workflow without ever returning to your desk. Here’s a concrete example using Copilot CLI:

1. Plan and scaffold with Copilot CLI: copilot plan "create a new REST API endpoint for user profiles". The agent begins generating a plan and creating files.
2. Use /remote on to send the session to the web or mobile.
3. Monitor progress on your phone while commuting — watch the agent read existing code, write new files, and run tests.
4. Steer the session with a follow-up: “Add input validation for the email field.”
5. Review the implementation plan and proposed changes directly in the web view.
6. Create and review a pull request right from your phone by triggering the agent to submit a PR via Copilot CLI or using the GitHub mobile interface.
7. Merge and move on once approved.

This removes the friction of switching surfaces — no need to be at your desk to push code or merge a PR.

Privacy by Default

Your sessions are private. Only you can see or access them. Remote control does not expose your code or session to other users. The feature operates within your existing GitHub Copilot security model, and no additional sharing settings are required.

Common Mistakes

• Not using the latest version: Remote control requires updated Copilot CLI or VS Code extension. If the /remote on command is not recognized, update your tools via your package manager (for CLI) or the VS Code marketplace.
• Forgetting to authenticate on mobile: On the GitHub Mobile app, ensure you are logged into the same GitHub account that owns the Copilot subscription. Otherwise, sessions won’t appear.
• Assuming remote control works for all session types: Currently, remote control is supported for agent-based sessions (those with planning and file modification capabilities). Ordinary chat-only sessions may not have full remote viewing. If your session doesn’t show up, try starting an agent session explicitly.
• Ignoring permission prompts: When steering remotely, you may need to approve file write or command execution permissions. If the session seems stuck, always check for pending permission requests in the remote view.
• Using an unsupported IDE: While remote control works with VS Code, JetBrains, and CLI, not all IDEs are supported. Check the latest GitHub Copilot documentation to confirm compatibility.

Summary

GitHub Copilot’s remote control feature transforms the way developers manage multi-agent workflows. By allowing you to start a session on one device and monitor, steer, and complete it from any device via /remote on, the tool eliminates the need to remain at your desk. Real-time visibility, mid-flight instruction changes, and full pull request lifecycle management are now possible from a web browser or mobile app. With privacy built in and no extra installation for existing users, remote control is a powerful step toward a truly multi-surface development platform.