Automation: the cloud coding agent

The Models & credits chapter closed Act 1: the careful shared-lib change shipped, orders-service consumes the new version, and the approval workflow is built. What’s left is a different kind of work entirely — a short backlog of follow-up issues on orders-service. A couple of validation edge cases the workflow surfaced. A small audit-report endpoint product asked for. None of it is hard, none of it is risky, and all of it is well enough scoped to write a sentence about.

That combination — low stakes, clear boundaries, your own app — is the profile from the modes chapter where you give Copilot the longest leash. Act 2 is where you give it the longest leash there is: you stop driving it yourself.

The same Copilot, a different surface

Back in the first chapter you met the idea that Copilot is one assistant wearing several surfaces — completions, chat, the modes. Here’s the surface that proves it. Everything you’ve done so far happened inside VS Code, with you watching. The cloud coding agent is the same Copilot, reading the same custom instructions and skills you already wrote — but it doesn’t run in your editor at all. It runs on GitHub, asynchronously, while you do something else, and it hands its work back as a pull request.

That changes the rhythm completely. In the IDE you and Copilot share a keyboard; the loop is tight and synchronous. On GitHub the loop is async: you describe a task, walk away, and come back to a finished proposal. The judgment from the modes chapter — match autonomy to stakes — is exactly what tells you this backlog is safe to run that way. Cheap to fix, easy to undo, no twelve downstream consumers. Perfect for delegation.

What this chapter covers

You’ll clear the backlog by handing each issue to the cloud agent and reviewing what comes back:

• Assign an issue to Copilot — the handful of places you can kick off async work — Assign · hand an issue to Copilot
• It proposes, you merge — the draft PR it opens, and iterating on it without leaving GitHub — Draft PRs · it proposes, you merge
• Customizing the agent’s environment — the one config file that shapes how it runs, and who’s allowed to turn it on — Environment · copilot-setup-steps.yml

By the end the follow-up backlog is clear, and you’ve seen the proposes-you-decide loop stretch all the way out to its async limit — the agent working on its own, you reviewing on yours.