← sources

Claude code is design to make you dump

source: reddit (reddit:ClaudeCode)published 2026-11-18 12:01ingested 2026-11-19 02:39

retained text

I've been writing code for the past 10 years, but a couple of months ago, I switched from Cursor to Claude Code after Cursor started pushing its own models and limiting usage of Claude models. I've been really enjoying Opus and Fable, but there's something I've noticed about how these tools are changing my coding habits.

Claude Code is set up in a way that makes reviewing code harder than it should be. You mostly get shown the diff, which makes it harder to explore other files and understand the broader context. On top of that, it runs deep review skills before committing, which often flag edge cases that don't necessarily need to be addressed. The agent then applies fixes automatically, sometimes making the code 10 times more complicated and harder to maintain without even asking for your approval.

Naturally, I've found myself becoming lazier and lazier with code reviews. And I suspect the same thing is happening with some of my colleagues.

A couple of days ago, after a release, we started getting weird reports from clients, and it became clear that parts of the codebase were becoming increasingly difficult to maintain. That pushed me to become much stricter about code reviews again, but honestly, I feel like these tools are designed in a way that works against that. Doing proper reviews takes more tokens, more time, and more effort than just trusting the agent.

I'm curious to hear what other engineers, especially those who don't consider themselves vibe coders, think about this. Have you noticed AI coding tools making you less disciplined about reviewing and understanding your own code?

document fetched 2026-11-19 00:41 · canonical https://reddit.com/r/ClaudeCode/comments/1xzdi6c/claude_code_is_design_to_make_you_dump