Woah, hold up. Kabyle has two approximants and two trills, not just a weeby tap. It's very out-of-place to have such an undiverse PoA given how many series of plosives and fricatives there are.

And now a rant about transliteration schemes.

As far as I can tell from their Wikipedia recordings, [ʕ] is barely audible, and contrast between [ħ] and [h] is ridiculously hard to hear. If you want those sounds, make them allophones, not distinct phonemes. Even if they get distinct graphemes, there should be some contextual rule about which is used, e.g. [ħ] in word-initial positions but [h] elsewhere. [ʕ] probably doesn't even deserve a grapheme, but could maybe work as the sound of lenition-in-progress as applied to a [ʁ] or [ɣ] consonant (as a counterpart to [x], since otherwise it seems there's plenty of voiced/unvoiced/aspirated-unvoiced contrast in the inventory.)

I think if this collablang is going to be safe from the level of autistic xenophilia that presumably brought about the aforementioned ten-tone lang, then attention really does need to be spent on naturalism. The original writing system of the Berber family, Tifinagh, is quite inexact about phonology—it was originally an Abjad with numerous variants and redundancies. Even today all three common orthographies (Latin, Neo-Tifinagh, and Arabic) have some amount of ambiguity in what a given letter represents. I can understand that, while constructing the language, it might be desirable to have an exact system of phonics1 but no everyday orthography has a perfect 1:1 correlation for more than a few minutes after its creation.