After a Month with Claude Code, I Went Back to Cursor
Honestly, just looking at autocomplete speed, Cursor still feels more convenient lol
When using Claude Code on large projects, the context drain is noticeable enough that I eventually went back to Cursor. That said, when quickly building new features, it was really convenient.
Verdict: Claude for small projects, Cursor for maintenance. Is it just me?
8 answers
Agreed—I also ended up coming back because Cursor absolutely dominates in autocomplete speed lol
I totally relate to it eating up context. When you use it in a big monorepo, you can feel it getting dumber the further you go.
Well, that's a bit... A one-month trial seems like too short a sample to draw conclusions from, no? And since you didn't mention the project size or language, generalizing seems like a stretch. I've used it for over three months and actually ended up switching over to Claude entirely. Early on I was stressed about context too, but once I organized rules in CLAUDE.md and split work into subagents, it made a clear difference.