Sometimes the terminal wont reach a natural termination point and stay alive, but will have encountered an issue. I have to manually cancel it, then it proceeds to understand what went wrong and fix it. Which is amazing.
But if the agent could periodically view the output and realize when it’s stuck and choose to terminate it automatically (even if this is an optional feature / setting), that would speed up development and complete the loop in an awesome way.
So many trial and error iterations could run without my manual interaction.
A very simple version of the feature could just be to terminate if no new output happens for a while.
Isnt a timeout already integrated? perhaps this should be a bug report. But overall i agree, sometimes the timeout isn’t fast enough as it waits for more.
I used recently console output and it stopped watching by itself. However i have also noticed that it sometimes didnt stop. So i think this needs at least clarified how it is processed and how it decides when to stop.
It also happens to me and to some coworkers. Is this being taken care of? In my case, it takes a very long time to solve/close the terminal, and most of the time, it gets stuck, forcing me to restart the console, close it manually, or pray to the gods to work…
We have solved several of such issues on 1.4.x and 1.5.x but if you still have them then kindly fill in a bug report as that helps find the cause or to reproduce it.
when I open the Cursor it all working well and suddenly it start to lose the task ended signal, so I notice it occur each time the OMZ make an update while the cursor is open and from that point it stop identify the command ended.
restart the cursor solved it in each case it occurred for me.
so make sure if you any terminal update or so, it might be your use case as well.
