1.3.7
Your own commands run
Type / and you get every command the session really has, your project’s own skills included — and they run. A machine you leave a workspace open on all day stays reachable, too.
New
- DesktopType / in the chat and every command the session really has is there, your project’s own skills included, and it runs. The list was ten fixed names before — half of them not real commands, and none of them yours.
Improved
- DesktopYour project’s own skills sort above everything a plugin brings, so the ones you wrote are the ones you see first. The list is shown whole and scrolls rather than stopping at the first dozen.
Fixed
- EngineA machine stays reachable however long you leave a workspace open on it. After a few hours it would go dead in the app while the machine itself was perfectly healthy, and only restarting Padar brought it back.
- EngineA workspace that has just restarted is reachable again straight away. It could come back unreachable and stay that way while the workspace itself was running perfectly, and reopening Padar fixed it only sometimes.
- DesktopClearing the conversation leaves a clean transcript. It used to read as though the agent had answered you with “(no content)”.
- DesktopThe agent stops looking busy after a command it answers instantly. Running /context or /usage used to leave it showing as working until you sent something else.