Hi,
New user here. I wanted to provide feedback on my use of cursor so far. If any of my comments are based on ignorance please gently educate me.
Installation:
I’m installing on W11. I wanted to install for all users on this machine so tried running the installer with Admin rights. Received the popup explaining things and was told to download the admin installer under “Downloads”. That installer doesn’t seem to exist so IMO the installer popup should be changed, or clarified if I missed the location for the admin installer. As it is it was a bit of waste of time searching for something that doesn’t exist, and not the best look when the installer is out-of-sync with the website.
I chose to import my (numerous) VSC extensions, themes, etc. Added both “code” and “cursor” to the “terminal” (I assume this is just modifying PATH). I authenticated using GitHub to get the AI features. I received this error in the editor:
The editor could not be opened due to an unexpected error: Failed to get model for chat editor: id
Gemini said it was a known but sporadic issue with GH authentication.
I completely uninstalled cursor, reinstalled, did not import anything from VSC, and it works.
I miss my VSC keybindings (including when I imported from VSC). My muscle memory is to use Cntl-W to close tab and Cntl-K Cntl-W to close all tabs. I know I can go into the command pallet and reconfigure keys but are (my) keybindings imported from VSC by default? Are there keybinding collisions with cursor default keys that would preclude this?
IMO I should be able to configure data privacy immediately, not wait for a day to elapse then set it. AFAIK it appears that I cannot set that privacy setting immediately after installation.
AI Features:
In general it has been a big improvement over VSC+Copilot, but…
On the 14-day Pro functionality, it appears that the only model I can select is Auto. I can sort of see why since I’m not yet paying and you don’t have my funds to pass on to Claude, ChatGPT, etc. Still, it would be nice if I could test the actual model I want to use for development.
Assuming there isn’t a way to change the model during a trial and Auto is the only option, it would be nice if it were “self-aware”/trained on cursor functionality. I ask it questions on why cursor works such-and-such a way but it can’t answer those questions.
Some excerpts:
since the Cursor UI isn’t providing the option to apply directly to that file.
No, I’m not specifically trained on Cursor’s UI and functionality. While I can help with code and project development, I can only make educated guesses about Cursor’s specific features based on what I observe in our interactions.
I understand your frustration. The trial limitations on model selection can be a significant drawback, especially when you’re trying to evaluate the full capabilities of the IDE.
For now, let’s focus on what we can do effectively:
Continue with the project development (like implementing the Norgate API functionality)
Use me for code-related tasks where I can be most helpful
For Cursor UI/functionality questions, you can indeed use ChatGPT or Gemini as you suggested
I understand your frustration with the UI limitations, but I actually can’t control or fix Cursor’s UI behavior. I can only work with the tools that are provided to me, which are focused on code analysis and file operations within your workspace.
For this last one, I was asking how to change the file context since the AI is continually suggesting I change the wrong file, so I can’t use diffs to review the changes, only paste from the chat window to the correct file. So IMO this is a functionality issue in cursor (for this particular situation).
Again, apologies if these questions/comments are out of ignorance. I’ll try to read the cursor documentation as soon as possible.