GitHub Copilot app for Beginners: Run several agents at once

GitHub Copilot, the AI coding assistant from GitHub, offers a feature that allows multiple agents to work on a project simultaneously without interfering with each other. This is akin to running several loads of laundry at once in separate machines, each with its own settings, without waiting for one to finish before starting another.

In the GitHub Copilot app, each AI agent session is a complete task that runs independently of the others. The app allows you to manage several sessions through a sessions view, where each session is represented by a card showing its title and progress. Each session operates on its own Git worktree, enabling parallel execution. This means you can start a new session at any time without disturbing the others.

Each session maintains its own context, allowing you to switch between them freely without losing track of where you left off. This reduces the time spent on context switching, enabling you to focus on decision-making rather than monitoring the agents. For instance, if you have three tasks for a repository like tailspin-toys - adding a funded sort feature, conducting an accessibility review, and running tests - you can ask Copilot to perform each task in separate sessions.

You can monitor the progress in the session view and review the results as they complete. You can also take breaks knowing that the tasks are happening independently. To experience the power of parallel agent sessions, try starting two small tasks at the same time in the GitHub Copilot app.