GitHub Copilot in JetBrains Air: Local, Cloud and Controls

GitHub announced on November 3, 2026 that Copilot is available in JetBrains Air through the Agent Client Protocol. Developers can select Copilot in Air’s agent picker, delegate local tasks, run cloud tasks and review several sessions alongside project context. The integration uses an existing Copilot subscription rather than JetBrains AI credits, but the account, administrator and governance boundaries differ between local and cloud work.

Local tasks combine Air orchestration with an existing Copilot account

A developer installs Air in a JetBrains IDE, selects GitHub Copilot and connects the Copilot account. Air can coordinate parallel sessions and present changes for review. The release does not say that every JetBrains product, Copilot feature or organization policy behaves identically, so teams should test their supported IDE and plan.

Running beside the ordinary GitHub Copilot plugin can also create overlapping assistance. Define which tool owns a task, keep changes on a branch and review the combined diff rather than assuming the interface preserves a single chain of reasoning.

Cloud tasks start clean, but still need explicit repository and network boundaries

Air Teams can launch Copilot in a Docker container on a fresh virtual machine. GitHub says traffic passes through a proxy and state does not carry between tasks. The agent can build, test and install packages without changing the local workstation.

Isolation reduces persistence and local-environment risk; it does not make repository content, package installation or outbound requests harmless. Use a least-privilege branch, protect secrets and inspect what the task downloaded or executed. A fresh machine is an execution boundary, not an approval decision.

Plan selection and usage visibility have important limits

Cloud use requires a JetBrains account in an Air Teams-enabled business organization, a Copilot subscription and administrator enablement. Air may check subscription or credit availability when the environment starts. GitHub says a user cannot choose whether an individual task consumes a personal or organization Copilot plan.

Air’s governance and analytics do not include Copilot usage, and Air administrators cannot set Copilot spending limits within Air. They can disable Copilot through policy. Organizations therefore need Copilot-side usage and policy review rather than treating Air as the complete administrative record.

A bounded pilot should measure completed code, not concurrent sessions

Start with a test repository and two repeatable tasks: one local and one cloud. Record account and policy state, files changed, packages installed, tests, human corrections, elapsed time and billed usage. Confirm that a failed session leaves no unexpected remote branch or credential behind.

For Canadian organizations handling regulated or client code, review data location, contractual terms and repository access before enabling cloud execution. The integration expands agent choice; it does not transfer responsibility for code quality, licences or security from the maintainer.

Sources and editorial method

We use the linked original statements and reports to distinguish documented facts from our analysis. Company announcements describe their authors’ plans and claims; they are not independent verification of future outcomes. This article is dated and will be revised if material evidence changes. Read our editorial standards and corrections process.

GitHub: Copilot is now available in JetBrains Air ↗ — Official November 3 release; ACP integration, local and cloud setup, isolation, account requirements and governance limitations.