GitHub Copilot remote control makes local agent sessions reviewable from mobile
GitHub says Copilot remote control is generally available for CLI sessions on github.com and GitHub Mobile, with VS Code and JetBrains support arriving as the workflow expands. The useful angle is not novelty on a phone; it is keeping long-running coding agents observable, steerable, and permission-gated when you leave the desk.
GitHub Copilot remote control lets developers start local Copilot sessions in the CLI or IDE and continue monitoring or steering them from github.com or GitHub Mobile. GitHub says CLI remote control is now generally available, while VS Code and JetBrains support are being introduced as part of the same multi-surface workflow. The practical value is session supervision: seeing plans, file reads, commands, permission requests, proposed changes, and pull-request steps without staying at one machine.
Key takeaways
- GitHub positions remote control as a way to move Copilot agent sessions across CLI, VS Code, web, mobile, and JetBrains surfaces.
- The announced CLI flow uses
/remote onso a local session can be monitored from github.com or GitHub Mobile. - GitHub says developers can send follow-up instructions, redirect scope, and approve or deny permission requests while a session is running.
- The feature matters most for long-running agentic coding work where visibility, review, and interruption handling are more important than raw generation speed.
- Teams should still verify repository permissions, data exposure, and pull-request review habits before treating mobile control as a production workflow.
Practical LinkLoot angle
Remote control changes how a coding-agent workflow can be managed: you can start a refactor at a workstation, step away, and still catch a bad plan before it becomes a large diff. That is useful for solo developers, but it is especially relevant for teams that already run multiple agent sessions in parallel and need an audit-friendly way to pause, steer, or reject risky actions.
| Workflow option | Best use | Limitation | Source |
|---|---|---|---|
Copilot CLI with /remote on | Start local terminal work and supervise it from web or mobile | Requires the Copilot CLI flow and account-level access | GitHub Blog |
| Copilot in VS Code or JetBrains | Keep agent work tied to the developer environment | GitHub describes remote support as being introduced, so teams should confirm availability in their IDE build | GitHub Blog |
| GitHub Copilot cloud agent | Delegate repository work to an agent-managed branch and review the diff later | Needs policy, access, and review controls for organization use | GitHub Docs |
A useful rollout pattern is to restrict remote-controlled sessions to low-risk repositories first, require small pull requests, and treat every agent command approval as a security decision. If a session asks to run migrations, touch secrets, change CI, or widen network access, do not approve from mobile just because the interface makes it easy.
What to verify before you act
Check whether your plan, Copilot tier, organization policy, and IDE version expose the exact remote-control surface you want to use. Confirm how private sessions are represented in your account, which repositories or local directories can be sent remote, and whether mobile approval flows preserve enough context for safe decisions. For teams, test the workflow on a non-production repository and inspect the final branch, commit history, and pull request before allowing agents to work on critical code.
Source check
The GitHub Blog announcement confirms the May 18, 2026 launch framing, CLI general availability on github.com and GitHub Mobile, the /remote on workflow, live monitoring, steering, permission approvals, pull-request review, and GitHub's statement that sessions are private by default. GitHub Copilot documentation corroborates the broader Copilot agent, cloud-agent, CLI, agent-management, MCP, access-management, and risk-mitigation surfaces that teams should check before deployment. The GitHub Copilot product page corroborates the broader positioning of Copilot across editor, terminal, GitHub, agents, custom agents, third-party agents, and MCP-connected workflows.
It is a Copilot workflow for starting a local agent session and monitoring or steering it from github.com or GitHub Mobile, with IDE surfaces also being introduced.
If you are building repeatable agent workflows around Copilot, compare this with the broader LinkLoot guide to AI agent tools and keep a checklist for permissions, repository scope, and review gates.