The only thing I pay for Cursor is the auto completion and tab functionality.
I’d like to know what’s the next plan for the auto completion and tab, and how far will Cursor push the tab and auto completion.
Can’t wait for the next tab release.
The only thing I pay for Cursor is the auto completion and tab functionality.
I’d like to know what’s the next plan for the auto completion and tab, and how far will Cursor push the tab and auto completion.
Can’t wait for the next tab release.
CPP is the main feature that keep me locked into cursor, most of other tasks I was using aider
.
The feature Im expecting and being very annoying in the email asking for them is the ability to use any other model in CPP, the only flaw of CPP is that is dumb for complex suggestions that turn it very annoying in the long run.
Same here. The only things that I pay are the things that really work:
Agents? They’re cool in theory, but rarely work in practice.
They work extremely well if you know how to prompt and have a good understanding of how the LLMs work. What limitations there are, etc.
Some things are quite hard to get right with AI, but there are a myriad of ways in which you can improve its performance to do almost anything.
Domain knowledge and coding skills will get you to at least 95%.
I think version 3.7 is actually a step back from the agent implementation in 3.5. If you don’t agree, there’s not much I can do to convince you.
.47 was horrible, so I switched back to .45 for a few months, then decided to try the .48 and see if there were improvements.
.48 has been incredible so far.
Agent works extremely well for me and appears to be to only eating crumbs from my usage based model. It completes basic tasks would take me a lot longer to code myself and is capable to completing much harder tasks that would take me even longer.
So far, no complains from .48.2 and agent.
Agree 100%.
I’ll also add that the Agent in Ask mode is incredibly useful for exploring and understanding existing codebases.