Repository navigation
Replies: 3 comments 10 replies
|
Thanks for asking, and typing in the main agent's prompt is fine. When a message is delivered to a seat, it's typed into that seat's prompt and submitted, so a half-written draft in the same prompt goes along with it. In 0.6.4 you can protect a seat you like to type into with the typing guard: While it's on, OpenRig holds all automatic input to that seat, even at an empty prompt, and keeps the messages and wakes in the seat's outbox instead. It's off by default, and it's an on/off switch rather than something that detects typing. A few things worth knowing:
— comm-replies@v-openrig-build, on behalf of @mvschwarz |
|
Not you doing it wrong, but it is expected given how delivery works — I checked the transport code ( The guards that exist protect the daemon's own submits, not text you type. So the practical rule: treat a driven pane as read-only for yourself. Compose outside it — draft the message elsewhere and hand it over with Whether the daemon should detect operator-composed text and defer/queue pod deliveries around it is a fair feature ask (it would need pane-diffing your keystrokes apart from delivered bytes — doable via the same capture-compare machinery, but new behavior). As of current main, no such guard exists, so composing in the pane is racing every inbound delivery by design. |
|
Further to this thread. When I chat with the project lead vis the openrig kernal operator it blocks some things until I tell the operator to do it, like github merges etc. is that correct? I Updated the OpenRig tonight. |
Uh oh!
There was an error while loading. Please reload this page.
I might just be doing something wrong... but sometimes I am writing a message on the main agent's prompt, and IN COMES A NEW MESSAGE from a pod and sends my half-written message....
Is this meant to happen? Or am I using it wrong and should not be writing directly in the main agent's prompt window?
All reactions