When constructing my cube, I try to design it in much the same way that R&D designs a modern magic set for limited. When it comes to the multi-colored cards and cards in general, I try to balance the colors/guilds/shards/wedges/lands to be even amounts so as to try to avoid a warped drafting environment where a particular color or color combination is favored. The difficult thing about balancing the multi-colored section comes with cards like guild mages, cards with kickers, cards with flashback, hybrid cards, off color activations, etc. While it is clear that a card like Armadillo cloak would be considered a multicolored card, what about a card like Forbidden Alchemy or Mystical Teachings? It requires a little bit of judgement of your part. I try to follow a couple rules
1. For cards whose casting cost is one color, but has a flashback cost/off color activation/off-color kicker I try and consider if the card is worth playing in a deck where you don't have the off-color mana. An example of this in my cube is Scholar of Athreos. In a W/B deck built for the long game, she is an allstar. As a 3 mana 1/4, she is pretty forgettable. Another example would be Dismantling Blow. 2W for Disenchant is strictly worse than Disenchant (which is in the cube already), but in a W/U deck, 4WU to disenchant and draw to cards is pretty respectable.
2. For hybrid mana costs, I try to consider what decks in the cube actually want to cast the creature. A couple examples from my cube are Rakdos Shred-Freak and Slippery Boggle. Red has plenty of aggressive hasty opportunities in the cube and RR for a 2/1 haste isn't a bad deal. Most of the red/black decks in the cube would play it, but might consider taking goblins instead. When it comes to Slippery Boggle, the card is there to be another Gladecover Scout for the W/G Aura deck. There isn't really any other reason that a blue deck would want a 1/1 hexproof.
3. For flashback costs, I consider whether a player would be happy with casting the spell once, or if you need to cast it twice to feel good about it. An example of the former would be Strangling Soot. Casting it just once is perfectly fine removal spell, being able to cast it a second time is excellent. An example of the latter is Reap the Seagraf. A 2B 2/2 is not good enough for the cube, but being able to flash it back, or get value off it if you self mill it, is relevant in Blue/Black which is Zombie tribal that benefits from self mill.
These distinctions become harder when you consider cards like the -scape cycles of familiars and apprentices from Invasion. I started the cube with both full cycles, but have dismantled them over time. Including cycles just for completion is a dangerous habit than can lead to some sub-par cards. As it stands I only have the Stormscape and Nighscape familiars left in the cube. Mostly because they support Esper control or Storm-ish decks. As far as the apprentices go, the green and blue ones are left in because they can be tappers. The black one is in because it is secretly a zombie and because U/B can take advantage of ETB effects while B/R aggressive decks could really use the first strike. I've seen both of those cycles categorized as wedge cards in other lists, but I don't think it is an apt categorization.
The most important thing for me is to use the multicolored cards to help give a player unfamiliar with the cube an idea of what the color combinations do. The U/W cards are there to show that artifacts are what is important in that color combination. Most of the B/G cards try to encourage graveyard play. This isn't a rule that I strictly follow (Consume Strength, for instance, has nothing to do with the graveyard. Beyond usually helping to put two of your opponents creatures into the graveyard). I'm more likely to cute two color cards that don't support specific decks because they don't help point to a particular archetype (this is most of the reason that Fire at Will, Putrid Leech, and Temporal Spring were cut in the most recent update).
TL;DR, I make it up as I go.