Early in the Cursor days (1.x versions?) I recall that there were ‘full’ and ‘compact’ options in Settings to govern agent display. In 2.6.11, agent display of executed commands has been abbreviated to a not-very-useful single line such as
Although one can expand each of these individually, I don’t see a way to expand them all by default anymore. Does this setting lurk somewhere?
@deanrie et. al, I’d like to put a finer point on this with respect to the urgency and risks of this change. Hiding sandbox output by default obscures the ability of the user to be an active participant in what the agent is doing. If a user isn’t shown the agent output, they can’t intervene when the system is going in circles or doing the wrong thing.
Specifically, I just had an experience in which the agent didn’t realize that we had a Python venv for tests, and because the output wasn’t shown, it went and installed a bunch of dependencies like pytest in the system Python installation. Had I been able to see the output, I would have stopped it before this occurred.
Please revert this change, and make showing agent sandbox output the default.
Hey @rsb
I think you’re looking for this setting under Cursor Settings > Agents > Inline Editing & Terminal > Collapse Aut-Run Commands.
That did the trick, thank you! I had a good look through Settings before posting, but there are just too many of them these days. This opinionated user feels that the collapse setting should be off by default. Something to consider for a future release.
This is odd. The Collapse Auto-Run Commands setting doesn’t seem to be honored correctly anymore. I double checked to make sure that it was unset (therefore, auto-run commands should not be collapsed):
but I’ve been seeing output like this all day in the agent…
This was working back on March 10th. I’m now on:
Version: 3.0.16
VSCode Version: 1.105.1
Commit: 475871d112608994deb2e3065dfb7c6b0baa0c50
Date: 2026-04-09T05:33:51.767Z
Layout: editor
Build Type: Stable
Release Track: Default
Electron: 39.8.1
Chromium: 142.0.7444.265
Node.js: 22.22.1
V8: 14.2.231.22-electron.0
OS: Linux x64 6.8.0-107-generic
(Request ID 0d282606-3490-4cdf-8dd9-e7e99e47e41c , just in case that helps).
Perhaps there’s been a regression?
There has been, unfortunately!
Hey @rsb,
The collapsed terminal output regression has been addressed in a recent Cursor update. Updating to the latest version should resolve this. Let me know if you’re still running into it!
Hey @mohitjain , not sure if I should file this as a new bug related to this item, or extend this one, but I have observed that the fixed behavior still isn’t optimal. By this, I mean that when the agent begins executing a command, the terminal isn’t expanded until the command produces output. This makes it hard to tell what’s going on without manually expanding the terminal output field – being able to see this is very important.
Let me know if you’d like me to open a new issue for this behavior.
Yes, this is a different behavior from the original regression. The “Collapse Auto-Run Commands” fix addressed the setting being completely ignored. What you’re describing now (terminal not expanding until output arrives, rather than when execution begins) is a separate issue.
I’d recommend opening a new thread for this as a bug report. Including a short screen recording showing the delay between command start and the terminal expanding would help the team scope and prioritize it.
