[dcl.init.list p3.10 The prvalue mentioned in the initialization of the reference abruptly appears without any specification

[dcl.init.list p3.10 says

Otherwise, if T is a reference type, a prvalue is generated. The prvalue initializes its result object by copy-list-initialization. The prvalue is then used to direct-initialize the reference.

The rule only says two things:

1. Generate a prvalue.
2. The effect of the prvalue(how it initializes the result object).

We didn't tell the story about how the prvalue was generated/ formed. Moreover, there is a case that the prvalue cannot be formed at all. It ought to be that we should associate the prvalue with the rules regarding copy-list-initialization. The suggestion might be:

Otherwise, if T is a reference type, an invented declaration U x = H; is introduced, where U is the referenced type and H is the initializer list; the program is ill-formed if the copy-list-initialization is ill-formed. The invented variable x is a prvalue. x is then used to direct-initialize the reference.

Start with [dcl.init.list] paragraph 3.10 and read the referenced copy-list-initialization rules. Done means the standard wording explains how the prvalue is formed and covers the case where that initialization is ill-formed, without leaving the proposed semantics implicit.