Yesterday, after receiving few updates on a bug fix that I requested and mentioned a month ago, I decided to revert back to a previous version to see if I could pinpoint where the change happened that caused the bug. For context, I primarily use Ask mode and in that mode the Apply button no longer diffs the lines specifically that I have mentioned and instead diffs my whole file which causes a severe slowdown especially as file size grows.
I was able to identify that the latest version of 1.5 didn’t have this issue and so I set about trying to use it for the day yesterday. This is where my grain of salt comes from. The features that have come out in the last few months since that version are actually really powerful. Having been a cursor user for more than a year now I have seen a lot change. The automatic context gathering really is a powerful feature and it does meaningfully produce better results. Cursor is still the best AI development IDE and I hope that is the direction that the team continues to go. You guys likely feel the pressure to compete with all the other tools that are getting the agentic flow running, but speaking frankly, I and many others use Cursor as a tool to bring AI into our flows granularly rather than agentically. I don’t want to discount agentic coding as a product direction, but there is a lot of competition in that space. Getting the granular stuff right, i.e. Cursor Tab feels like Cursors biggest win and what makes me want to stay to begin with. I hope that those who want to be granular won’t be forgotten as you dive more into the agentic space.
With all of that said, I just want to say that is still feel that Cursor is the best tool and I am grateful for the iterations that have been made. We make a fuss because we care.