GitHub has introduced the GitHub Copilot app, a desktop “control centre” for directing multiple AI coding agents at once.
The app is built around a single 'My Work' view that shows active agent sessions, issues, pull requests and background automations across a developer’s connected repositories.
Each agent session runs in its own isolated copy of a branch so parallel agents don’t interfere with each other, and a feature called Agent Merge can carry a pull request through continuous-integration checks, required reviewers and merge, with the developer deciding how much it automates.
The app is available in technical preview for existing Copilot Pro, Pro+, Business and Enterprise users.
Alongside the app, GitHub announced a broader set of agent tooling.
Canvases are shared work surfaces where people and agents operate on the same plan, pull request or terminal session. Cloud and local sandboxes give agents a bounded, policy-controlled place to run and test code, either on the developer’s machine or in an ephemeral GitHub-hosted environment.
Copilot code review gains a “medium tier” that routes higher-impact pull requests to a stronger reasoning model, and now works natively in Azure DevOps.
The GitHub Copilot SDK is now generally available across six languages including Python, Go and Java, and the Copilot command-line tool adds a redesigned interface, on-device voice input and scheduled tasks.
A set of partner agent apps - from vendors including LaunchDarkly, Sonar, PagerDuty and more - can also plug into Copilot.
.jpeg&h=142&w=230&c=1&s=1)
.jpg&h=142&w=230&c=1&s=1)
.jpg&w=100&c=1&s=0){kind=logo}
.jpg&w=100&c=1&s=0)
.jpg&w=100&c=1&s=0)
.jpg&w=100&c=1&s=0)
