
Cyclomatic complexity, method length, naming conventions: every one of those metrics was a workaround for human cognitive limits. Agents don't have those limits. So what survives, and what gets renegotiated? @docondev shares reflections on craft:
link.testdouble.com/fdff50

English














