This is the second post in the series on Orbit. In the first one I covered what it is and how to get started. This one is about the menu.
An agent stops and it's your call
When you have several agents working, each on its own thing, sooner or later one stops and needs you, and you have to decide whether to let it keep going, correct it, stop it or step in and finish it yourself.
Orbit lets you do all of that, but there are a lot of options and you're not going to memorize them. I built it. I still forgot half of them.
The menu tells you what you can do
So I gave it a menu. You go to the task, open it, and you get what you can do with it right now.
The menu also tells you what you can't do
What you can't do shows up too, greyed out and with the reason: if the agent isn't working, you can't pause it, and the menu tells you so instead of staying quiet. That way you're not stuck pressing something that does nothing, wondering if it broke.
The board has its own menu
Anything that isn't about one task in particular, like flows, rules or settings, goes in a separate menu. That one also tells you when something doesn't apply: if you try to open Orbit from inside Orbit, it answers that you're already there.
Inside a task, the menu tells you where everything is
When you open a task to see what the agent did, the menu also shows you what's in each tab: the diff, what it cost, what each phase was asked. You see what it did and decide what to do from the same place.
The video shows all of this on an example repository, and getorbit.sh has one for every feature.