Cursor vs JetBrains for a team that still ships .NET — what did you standardise on?
Half the team loves inline completions, the other half lives in refactor-heavy workflows. I am trying to avoid a holy war and still pick a default stack.
What decision criteria actually worked in practice?
15 replies
We picked Cursor for greenfield TypeScript and kept Rider for the big .NET monolith — shared formatting rules bridge the gap.
Standardising on one IDE failed; standardising on formatters, analyzers, and PR templates succeeded.
Junior devs onboarded faster with richer inline hints; seniors wanted deeper static analysis — we bought both licences.
The real fight was not the editor — it was agreeing on a single solution layout and test command everyone can run headless.
We measured time-to-first-PR instead of editor preference — surprising how little difference there was once docs were good.
Remote pairing pushed us toward VS Code family so screen sharing and Liveshare just work.
Compliance blocked cloud-connected assistants for one client; we maintain an offline profile for that engagement.
We wrote a one-page 'when to reach for AI assist' guide so people use it for boilerplate, not security-sensitive refactors.
JetBrains structural search still wins for me when I need to rename a concept that spans twenty files.
Cursor shined on greenfield APIs where tests were thin — the assistant guessed sensible happy paths we later hardened.
Budget-wise, mixing licences was cheaper than losing a hire who refused to leave their preferred environment.
We run weekly internal demos of real PRs so people see how others integrate suggestions without rubber-stamping them.
The team that ignored code review discipline saw more AI-induced regressions — tooling cannot replace culture.
I wish we had invested earlier in pre-commit hooks; they catch sloppy merges regardless of which editor produced them.
Pick defaults for CI and docs, stay flexible on editors, and measure defects per thousand lines instead of arguing taste.
Join the conversation.
Log in to reply