GitHub Copilot’s Linear agent becomes generally available

Issue trackers just became another place you can fire a coding agent without opening an IDE. GitHub’s changelog says the Copilot cloud agent for Linear is generally available: assign a Linear issue to Copilot, and it works asynchronously toward a draft pull request you still have to review.

That is the product promise in one loop. Ticket in. Branch and draft PR out. Human on the merge button.

Assigning a Linear issue to Copilot end-to-end

The flow is deliberately boring, which is the point. You assign the Linear issue to Copilot. The agent reads the issue contents, opens a draft PR, and keeps working in the background. Progress updates stream into the Linear activity timeline so the rest of the team sees status without polling GitHub Actions logs. When the work lands, Copilot requests a pull request review from you.

Setup is an org-level install, not a personal toy toggle. GitHub says you install the GitHub Copilot for Linear app from the Marketplace, and you need GitHub organization owner permissions plus Linear workspace admin rights. Teams that already live in Linear can keep triage where it is and only jump to GitHub when the draft PR shows up.

Ephemeral Actions environments and draft PRs

Copilot does not SSH into a developer laptop. GitHub says each run gets its own ephemeral development environment powered by GitHub Actions. That isolates the agent session, ties compute to Actions, and keeps the artifact of record where engineering already reviews code: a draft PR.

Draft is doing real work in that sentence. The agent is autonomous and asynchronous, but GA still ends in human review, not silent merge to main. For security and compliance teams, the interesting surface is the Actions environment and the permissions the app holds — same class of questions you already ask about CI runners and bot accounts, now attached to issue assignment.

GA extras: model pick, custom agents, steering

GA adds control knobs inside Linear, not just a binary on/off. Per the changelog, you can choose which model Copilot uses for a task, point it at a custom agent from your repository to match team workflows, and set the base and working branch so the PR targets the right line of development. You can also steer a session by mentioning Copilot in a comment with new instructions while it works.

Those settings can be per issue, or applied across a workspace or team with Linear agent guidance. That is how GitHub is trying to keep “assign to Copilot” from becoming a black box that ignores your branching model and house style.

Which Copilot plans get access

Copilot cloud agent for Linear ships with Copilot Pro, Pro+, Business, and Enterprise. Free-tier curiosity will bounce. Docs for agent guidance, models, custom agents, and branches sit in GitHub’s Linear integration documentation linked from the changelog.

Analysis: this is GitHub planting an autonomous coding loop inside a tool product managers and eng managers already open all day. Linear becomes the assignment surface; Actions becomes the sandbox; the draft PR remains the accountability gate. The upside is fewer context switches on well-scoped tickets. The downside is familiar agent risk — noisy PRs, over-trusted bots, and Actions spend — now one assignee click away. If your issue hygiene is sharp, GA is leverage. If your backlog is vague, you just automated thrash.