GitHub Copilot joins JetBrains Air, with cloud tasks billed to your Copilot plan

GitHub, Microsoft’s developer platform, says GitHub Copilot is now available in JetBrains Air. Developers can pick Copilot as the agent, locally or in the cloud, and cloud tasks draw on a Copilot plan rather than JetBrains AI credits.

What you need to run Copilot in Air

The integration uses the Agent Client Protocol (ACP). For local tasks, according to GitHub, a developer installs Air in a JetBrains IDE, selects GitHub Copilot in the agent picker and connects an existing Copilot account. Copilot can be used alongside the separate GitHub Copilot plugin, and GitHub says users can choose whichever workflow suits them.

Cloud tasks have stricter requirements. They need a JetBrains account in an Air Teams-enabled business organization, a Copilot subscription, and Copilot switched on by the Air administrator. There are two routes. In the browser, the user connects GitHub under Settings > VCS Providers, chooses a repository and remote branch, selects Copilot and sends a prompt. In the IDE, the Air plugin offers a Cloud option in the environment selector.

GitHub describes the cloud environment as a Docker container on a fresh virtual machine. All traffic goes through a proxy and no state carries over between tasks, so Copilot can build, test and install packages without touching the developer’s local setup.

Gaps in billing, visibility and control

The changelog lists several limits that matter to teams. A task cannot be assigned to a personal or an organization Copilot plan; the user does not get that choice per task. Connecting Copilot to Air requests no Copilot permissions, so Air may only check subscription or credit availability when the cloud environment starts. If either is missing, the task fails and an error appears in the session.

Governance is thin for now. Air’s analytics and governance features do not include Copilot usage, and administrators can disable Copilot with a policy but cannot set Copilot spending limits inside Air. For companies that already track agent spend centrally, that is a visible hole. It echoes the billing questions around Copilot Automations, which are billed to the repo owner.

A shared protocol lets one editor host rival agents

The technical point is ACP. It is an open standard that lets coding agents plug into an editor or IDE, instead of each vendor building its own integration. According to coverage of Air’s March 2026 public preview by SD Times and others, Air launched with Codex, Claude Agent, Gemini CLI and JetBrains’ own Junie, and JetBrains planned to add more through an ACP agent registry. Copilot now sits in the same picker as direct competitors from OpenAI, Anthropic and Google.

That changes who owns the relationship. At the preview, as reported by ADT Magazine, Air required a JetBrains AI Pro or Ultimate subscription or the developer’s own API keys, and cloud execution was still a tech preview. Now a developer who already pays for Copilot can reuse that subscription for both local and cloud work. JetBrains bills nothing for those cloud tasks, and GitHub keeps the usage.

The model also reflects a wider shift toward agents that run in parallel. GitHub says Air lets users track several agent sessions, use project context and IDE tools, and review changes beside their code. How well Copilot performs against the other agents in that picker is not addressed in the source, and GitHub offers no benchmarks or pricing details beyond saying tasks use the existing plan. Which Copilot tiers are eligible, how many credits a cloud task consumes and when the feature reaches all Air users are also unstated; the changelog points to a comparison and FAQ page for detail.

The same protocol thinking underlies tool connections for agents, as explained in our guide to the Model Context Protocol, though ACP concerns the link between editor and agent rather than between agent and data.

GitHub invites feedback through in-product channels, a feedback repository and a private survey, which suggests the integration is still being shaped. The Air plugin for JetBrains IDEs is itself labeled alpha on the JetBrains marketplace, according to the links in the changelog.