Speed versus safety tradeoff
Your team has gotten fast — agents let two engineers ship what used to take five — and the org is thrilled. You're the staff engineer who sees the other side: PR volume has tripled, reviews are getting rubber-stamped, and last week an agent-written change passed review and took down a downstream service for 20 minutes. Leadership doesn't want to slow down. How do you risk-manage a team that ships fast with AI without killing the speed, and what's the failure mode you're most worried about that nobody else is naming?
route_agent_prs(lines_changed: list[int], touches_shared_boundary: list[bool], behind_flag: list[bool]) → list[str][[80,500],[false,false],[true,true]]out["auto_merge","split"][[120,120],[true,true],[false,true]]out["flag_then_canary","canary"][[200],[false],[false]]out["human_review"]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.