Quality problem

Ideas Feedback
,
1

I just came here to report a bug where subagents finishing caused the agent to restart without permissions. A pretty bad bug affecting very basic functionality that has been released for weeks/months.

However I can’t report it because of a hanging upload on the bug report forum post form.

Guys I’m sorry but at the point that even reporting a bug is broken, I think you have a severe quality/testing problem.

I wanted to start a discussion about this because while I respect the immense amount of work and effort the Cursor team is putting in, I think this is a sign of too much pressure. There is a rush to slam out a lot of features and products that may or may not be useful such as iOS app, agent window, etc. And I’ve enjoyed both of those! However they won’t make issues such as the original bug I mentioned “okay”.

How can we best make our preferences as customers clear to the team and reinforce that the constant bugs and broken updates are not okay? Why is this pervasive quality issue so bad, why is it not getting any better? Is it just considered acceptable to break prod every week now?

3

Cursor has uninstalled itself from my Windows machine after every update and still does so. I’ve had to download/reinstall over 20 times over a space of a year.

There’s a clear caste system in play. If you’re a Windows user or not perceived to be a high value user there’s no reply or response.

I don’t blame the Cursor team, I guess they have to triage and go after high leverage tasks and they are chasing the frontier of what’s possible. Quality assurance, forget about it.

5

This resonates with a broader issue I’ve been seeing: as coding agents become more autonomous, the cost of being confidently wrong increases. More features aren’t necessarily the next leap; reliable reasoning, verification, and knowing when an agent doesn’t have enough evidence to act are becoming just as important.