Practitioner essay: rejecting agent code that works but does not fit

Category: Agentic coding
Status: discussion
Sources: vinibrasil.com, discussion
Summary: A practitioner essay argues for rejecting agent-generated code that passes tests but ignores the codebase's conventions, abstractions, or intent, on the grounds that code is read and maintained far more than it is written. The post reached the HN front page (124 points).
Why it matters: It frames a recurring review tension as coding agents produce more functionally correct but stylistically divergent diffs that still cost maintenance effort.