AI code review rejection
A respected senior reviewer rejects your PR with the single comment: "This is AI slop — rewrite it by hand." The change is a Rust module for parsing a binary wire format; it's verbose and over-commented in places, and one helper is more generic than it needs to be, but it's correct, tested, and you understand all of it. The reviewer's stance is hardening into a blanket "no AI-generated code" position. How do you handle this — both the immediate PR and the underlying team disagreement — without escalating into a tooling holy war?
triage_slop_review(line_count: int, baseline_lines: int, comment_ratio_pct: int, unused_generic_params: int, tests_pass: bool, provenance_objection: bool) → list[str][180,120,55,1,true,true]out["over-commented","over-generic-helper"][100,100,10,0,true,true]out["ask-for-specifics"][100,100,10,0,true,false]out[]Treat the AI’s output as a draft to verify, not an answer to trust. Name the specific flaw and the input that triggers it, say how you’d catch it (tests, edge cases, reading critically), and how you’d re-prompt or decompose to get it right.
Vibe & agentic: describe the solution in plain language (or narrate it) and the coach grades your approach.