SWE-Gate: Why Passing Tests Isn't Enough for Agent-Generated Code

How review-derived acceptance constraints expose the gap between functional correctness and production readiness in coding agents.