GitHub Copilot app for Beginners: Run several agents at once

GitHub published a beginner-oriented post explaining how the GitHub Copilot app supports running several agent sessions at once. According to the post, running multiple AI agents on the same project can seem chaotic, but in the Copilot app the agents work separately and do not interfere with each other.

An agent session is described as a task given to Copilot from start to finish. In the app, several sessions can be managed through the sessions view, where each card shows its title and how far along the agent is in completing its task. Sessions run independently, so starting a new one does not disturb those already running.

The post states that each session in the app can run on its own Git worktree, and that because each session is isolated they can run in parallel at the same time. Each session also keeps its own context, allowing users to switch between them and pick up where they left off without re-explaining what they were doing.

The post uses an example repository, tailspin-toys, with three intended tasks: add funded sort, perform an accessibility review, and run tests. It describes asking Copilot to build the funded sort feature, opening a new session to request an accessibility review as that begins, and starting a third session to prompt Copilot to run tests. The sessions can be tracked in the session view and results reviewed as they finish, or the user can step away while the work proceeds.

GitHub suggests starting two small tasks at the same time in the GitHub Copilot app as a low-stakes way to watch tasks progress independently, and links to start using the app.