I am generally in favour of adding complexes in a flexible and utilitarian manner, but not on a personal whim. Yes someone obviously has a certain amount of knowledge if they are suggesting a complex, but they probably don’t have universal knowledge of the taxa - so consultation is important.
As far as monophyly goes, I’ve always taken the explanation in the curator guide to mean that it is just insisting that they must have the same taxonomic parent. (i.e. don’t try to make a complex between a species in Xylota and species in Chalcosyrphus just because they look similar.
As examples of the above: I helped organise Platycheirus taxonomy a while back, and one of the things I wanted to achieve was to replicate the species groups used by the UK recording scheme. But I had to check with workers in other countries that it wouldn’t tread on their toes, figure out if there were non-UK species that also needed to be included in certain of the groups, and ensure that they were compatible with the wider literature-based taxonomy we were trying to implement (oh, and pleeease do vote for this if you agree btw! https://forum.inaturalist.org/t/create-a-new-taxon-rank-for-inaturalist-the-species-group/14543 )
I was able to implement most of the groups - and I could find literature references to tell me what extra non-UK species should be included in them. But there was one I could not implement with the UK definition because its members were spread between different subgenera (it is a group of convenience for recording, but not natural), and one other that I had to include a species that the UK scheme puts in a different group. Basically, I had to accept a more internationally applicable definition. This is the sort of process I think the curator guide is calling for.
I agree with those who don’t think it is necessary for a paper to exist explicitly and exactly saying ‘There is such-and-such a complex’, but there should be an indication in the literature that these particular species constititute a recognised group of similar species. This I think is simply to avoid the taxonomic anarchy of people making up whatever they want, and including whatever they want in it. We are trying to be followers, not inventors of taxonomy. Having said that, there are some instances where a little exercise of judgement ought to be excused when the literature answers most, but not quite all of the relevant questions.