Great list! #1 and #5 really resonate—treating AI like a teammate and giving it context pays off. Also, #6 is key: always read every line, because AI can produce plausible but subtly wrong code.