S>> why do you want to use a tube with a grid top cap?
K> for aspect reason
In Maine I think we would say "for appearance".
"Aspect" is good English but tends to be used more technically. "Aspect Ratio" is length-to-width proportion, important in structural problems. "The financial aspect of the project" does mean "the appearance of the money" {does the money make sense?}, but often means the project is in financial trouble (the money situation is ugly).
> big voltage on the grid of a CF.
Yes, but usually a lot of resistance.
Mats' posted plan will be over 50 Megs to about 100V. A 2uA shock. Probably won't even feel it. Almost safe enough for open-heart surgery.
Another plan I saw here today had a 5.6Meg+1.5Meg divider. 1.2Meg to about 60V, 50uA shock. Safe enough through skin.
He "could" direct-couple out of a tube with 100K plate resistor to 300V. That's 3mA, which is actually "legal" (shock-protection GFIC/RCB devices trip at higher level) but enough to sting, and worst-case enough to stun a weak heart.
It needs consideration. It is easy to design it non-dangerous.
The conventional "hat" for grid-caps is a metal cup which is a tight fit on the top-cap ring. IIRC, there is a US-size copper tube fitting which is close-enough after a little saw and hammer. The metal tube shell is grounded via pin 1. The cap adds shielding on the grid lead, essential for the microphone input stage (not so much on a power tube driver).
The 807 power tube top-cap is the big danger. 400 Volts exposed in an open cabinet can send you to Heaven without a ticket.
I was going to say all GRID-cap tubes are metal, but I thought of an exception. 1D8GT is a diode-triode-pentode in glass. Top-cap is pentode grid. Pentode is a "power" pentode (0.2W), needs 5 to 9V of drive, so the top cap probably does not need shielding. (Especially since the main use was battery beach-radios, far away from hum sources.)