Often the output is cut or not provided at all even if the command run successfully.
A simple command like Start-Process "https://www.google.com/"
might fail according to the cursor agent.
I use powershell.
Often the output is cut or not provided at all even if the command run successfully.
A simple command like Start-Process "https://www.google.com/"
might fail according to the cursor agent.
I use powershell.
I sometimes see this myself. A command might seemingly be truncated, then the AI will see the command failed and try again in a shorter version or alternate way. But the original command the AI asked for as valid, it just didn’t seem to make it into the terminal window correctly. Not sure if there’s any useful way I can bug report those to the team with useful behind-the-scenes data for debugging.
Side note, I did install PowerShell 7.5 so “&&” would work. The AIs love to try using && even though they know I was on PowerShell and Windows default PowerShell can’t do &&.
I’m having a situation where I retry a command multiple times and I’m wondering if this is wasting my quick requests
There is a thread that suggest PowerShell 7 and a couple of IDE settings; but they don’t really help.;
The main issue just seems to persistently be that the agent moves on while the console is still running or then can’t see the output.