All orbital mechanics in ED is simplified down to a nested series of two-body problems. That's because more-than-two-body problems are dynamically unstable, even chaotic, and so cannot be predicted by the simple Newtonian modelling employed by the Stellar Forge. And if there's one thing that a procedurally-generated universe like ED absolutely needs, it's 100% reliability in its prediction model. If you introduce chaotic orbits into the ED prediction model, then two instances of the same solar system at the same time could look very different to the players in those instances.
Every object in the ED universe is either completely motionless (relative to another nearby body), or is orbiting one other thing - even if that "one other thing" is actually a barycentre of two other objects orbiting each other. Thus, in the example posted by Ozzie above, moons 7G and 7H are orbiting each other, while moon 7I is orbiting the barycentre of 7G and 7H, while the barycentre of the entire (7G-7H)-7I) system is orbiting the brown dwarf 7, which in turn orbits the system's primary star. None of these moons are "feeling" the gravity of any other object, other than the one they are deemed to be orbiting.
The closest thing we have seen to more complex orbital mechanics is the rare occurrence of Trojan worlds situated in a larger world's L4 or L5 Lagrange point. And these are stable simply because once the Stellar Forge system generator decides to make Trojans, then the Forge's Newtonian dynamics prediction model can keep the Trojan configuration going indefinitely without the need to actually invoke n-body gravity calculations.