This is really about a style of writing and comprehension that has caused, I think, much confusion. It is in this obscure place because I first tracked down the problem in this context, indeed in my own writing.

There are quite a few adjectives that we apply to keys. There are several kinds of adjectives as well and these kinds, by and large, have no names. I am confident that what we write about Keykos requires understanding these distinctions. This does not mean that we need names for these kinds of adjectives, however.

Here are some adjectives that apply to the keys (capabilities) of Keykos:

The Keykos savvy reader will already object to this list as comparing incomparable concepts but that is just the point I am raising — the new reader has grasped one kind of distinction between keys and expects the next adjective to be the same sort of distinction. How is the reader to discover that “factory key” is not a type of key that the kernel understands as it understands whether a key is a start key? The reader consulting the material as reference has no central place to go to learn of the distinctions between the distinctions.

Start keys and Resume keys are two key categories in a taxonomy known to the kernel. Collectively they are gate keys.
A brand is a role that any key can play. It might be better to speak of the brand slot of a domain and the key from the brand slot and thus avoid characterizing the key with an adjective. Keeper keys are another example — perhaps we should say that the key in the domain’s keeper slot is invoked when the domain encounters an invalid instruction. This way of speaking entirely eliminates one adjective for keys. I don’t expect to rewrite all the Keykos literature, however.

A requestor’s key is a start key to a domain that obeys factory code. This sort of adjective identifies the user code that the domain manifestly obeys, and thus defines the keys behavior. Other start keys to factories, such as builder’s key cause different parts of the factory code to interpret the message.

Humans make more kinds of distinctions than they have names for. Pushing names for these kinds onto the reader may not be productive. It might suffice to use canned phrases such as in

A Resume key is a key that designates a waiting domain. It is a variety of gate key and comes, itself in three sub-varieties: return, fault and restart. All of these are distinctions made by the kernel.
Additional confusion comes from the perverse perspectives that deliberately blur the boundary between kernel and other concepts. This perspective is, alas, a vital design tool. Indeed some Keykos programmers can work without knowing which keys are gate keys and which are primary. Those programmers still need crisp categories such as prompt.

Part of the strength of these systems stems from their automorphisms which play havoc with terminology.

I don’t know the answer.