Just shipped framework detection in cursor-lint — auto-detects your stack and suggests matching rule presets. Different rules for different setups. github.com/nedcodes-ok...

Sign in to view linked content

Comments

[Ben Halpern]

Nice project!

[Ned C]

Thanks Ben! Means a lot coming from you.

[Guilherme Zaia]

Curated lint rules are helpful, but Cursor's real bottleneck isn't linting—it's context window management. How do you handle false positives when the AI misinterprets your ruleset mid-refactor?

[Ned C]

This was something I was concerned about myself; I just finished a test where I loaded 50 rules at once and ran the same refactor 18 times.

Surprisingly, it hit 100% compliance even with 50 rules loaded. The "misinterpretation" usually happens when rules lack alwaysApply: true or have vague instructions, rather than the model forgetting them.

I just wrote up the full data here if you're curious: dev.to/nedcodes/i-loaded-50-rules-...