In the latest update all shell commands being executed are beingh hidden, obscuring what the agent is doing and allowing for security issues to whiz by if a setting is wrong. This should be a setting at a minimum with default safety.
Steps to Reproduce
Agent runs commands, cant see what its doing or its output.
Hey, thanks for the report. Could you share a screenshot of what you’re seeing? I want to confirm if the commands are fully invisible, or if they’re just collapsed and you can click to expand them.
Recent updates added collapsible output for terminal commands in the chat panel. They might look minimized, but you should still be able to expand them by clicking.
If they’re completely gone with no way to view them, that’s a different issue.
Also, are you using auto-run for terminal commands, or manual approval? And does restarting Cursor change anything?
Let me know. This will help narrow down what’s going on.
They are collapsed in a new UX pattern. Im in manual approval mode. It would be very nice indeed to have at least a settings option to say “auto collapse approved commands” on/off to preserve the old behavior
This is an absolutely horrible design decision that I can’t for the life of me figure out why you went with.
Users of Cursor shouldn’t be forced to click a button every time they want to see the output of the commands the agent runs. What were you thinking? And there is no way to disable it either? Wow.
