> LM384 has an absolute maximum input voltage of +/- 0.5V.
By golly, it does!
As you imply, we'd never get near there in "normal" use. The output would be clipped, totally distorted.
A peek at the internals suggests that at -0.6V the input base-collector junction will forward-bias and conduct big. If unlimited, it would blow. 0.5V gives a little safety margin.
Not stated on LM384 sheet: "nearly all" analog pins can take 10mA without damage. They need to be this rugged to shrug-off static electric discharge in handling.
So we could wire to 100V as long as we put 10K in series. And 10K against the internal 150K is hardly-any signal loss in normal use.
A "proper" 12AX7 stage runs on 300V. We could go 33K series with trivial loss of signal. However a 12AX7 stage only eats about 1.5mA. It's all 100K load and 60K internal impedance. A worst-case could be 300V/60K or 5mA. A 12AX7 stage is not likely to melt the input of an LM384.
300V supply on top of 6/12V for heater and 9V-18V supply for LM384 is getting to be more supplies than amplifier-stuff. Also 300V in classrooms sounds dangerous. And we don't really need "maximum performance" to flog a little hi-gain chip. It is tempting to starve the tube with perhaps one 12V supply for LM384, heater, and plate.
But tubes do poorly at low voltages. Their output declines much faster than the voltage, and their happy-bias becomes unpredictable and erratic. An old rule of thumb is to select Mu < Vsupply. 12AX7 Mu=100 can work dandy with 100V supply (table lists 90V conditions for ACDC radio use), but may punk-out at lower supplies. 12AU7 Mu=20 is a better bet. Happy 12AU7 gives gain of 15, which may decline to gain of 10 at 12V supply. LM384 at 12V supply is 4Vrms max output, gain of 50. Gain of 10*50 is 500. 4V/500 is 8mV input sensitivity. This is plenty of gain for guitar, we more often aim at 20mV for all-around use. We can take that difference in a volume control between the tube and the chip.