The best small UX upgrade you can give a serial-controllable sketch costs nothing: a help menu. Type "?" and the chip prints what it can do. Suddenly the sketch is discoverable. Anyone, including yourself in three months, can connect and figure out the protocol.
The cost is one extra command
You're already parsing commands. Add one more: when the input is "?" or "HELP", print a list of the others. The Arduino's F() macro lets you keep those help strings in Flash so they don't eat SRAM. A multi-line help message becomes Serial.println(F("ON — turn LED on")); Serial.println(F("OFF — turn LED off")); ....
The conceptual win is bigger than the code. By writing the help menu, you've forced yourself to inventory your own commands. Often you'll notice that two commands really should have been one, or that your naming is inconsistent ("LED_ON" vs "off"). The help menu is also a design tool.
This is a REPL
You have just built a REPL — read, evaluate, print, loop. The Lisp interpreter from 1958 worked exactly like this. So do node, python, the MongoDB shell, and Cisco's IOS console. Your tiny Arduino sketch is in good company. Once you have a REPL, adding scripted control, automation, or remote operation is a small step from where you are.