It looks like the hitch needs to be tied, then thru the dogbone, and have stopper knots tied, somewhat making it where a person will have to fight to get the stopper knots untied in order to remove the hitch from the rope.
As is, the stopper knots on the hitch cord are pulling up on the dogbone, guided by the slot, forcing the dogbone against the rope, and rope against the body.
Is it possibly to somehow change the top slot angle (or something) and/or move more force to the bottom biner/ rope area/ body by threading the dogbone and tying scaffold knots on the lower biner on each side of the body? I know that threading the dogbone to the biner will have a lot different vector force on that top dogbone (much less force).
Where is the force being split on the latest configuration between the dogbone and biner, to your guestamation? 50/50, 60/40, 40/60?
Another thing that I wonder about, that seems likely not to be a problem, but at least something to consider/ rule out, is whether there is a high pinch point force on the rope from the dogbone and/ or biner. Seems like the force is spread out between the hitch, dogbone, and biner pretty well.
With a Gibbs ascender, where there is one pinch point, a shockload from a fall can part/ break the rope at a significantly lower force than the typically breaking strength, IIRC. Seems like its in the 2XXX pound range, maybe, IIRC. For this reason, prusiks are preferred to connect MA system, I believe, rather than a Gibbs with quick release pin.