Steering vs Queueing in GitHub Copilot Chat
GitHub Copilot Chat has two ways to send a message while the AI is working: steering and queueing. The one you pick changes whether the AI stops to listen or finishes first.
Steering
Steering happens when you send a message while the AI is busy. The AI is in the middle of thinking and writing code, and you press Enter with a new message. The AI drops what it’s doing and reads your new message right away. It gets interrupted.
This is the default behavior in VS Code.
Queueing
Queueing happens when you send a message while the AI is busy, but you press Alt+Enter instead of Enter. Your message gets queued. The AI finishes its current task first, then picks up the next one. No interruption.
Queueing is the one I use. I don’t usually watch what the AI is doing in real time. I just want to hand it tasks and come back later. With queueing, I can send several messages at once and let the AI work through them in order, one task after another.
Steering gets in the way here. If I send a message mid-task, it interrupts the agent, even when I didn’t mean to. Queueing avoids that, which is why it’s my default.
Summary
| Key | Behavior | AI Handling |
|---|---|---|
| Enter | Steering | Interrupted, reads new message now |
| Alt+Enter | Queueing | Finishes current task, then reads next message |