I have a couple of questions/observations about what you're trying to accomplish.
(1) Why do you want to combine the hitboxes and ragdoll shapes? Those are two very different needs, and it seems to me that trying to double up two unrelated pieces of functionality like that is likely to cause you lots of headaches, both to get it working initially and to maintain it later.
(2) Why is the specific triangle that a ray hits important? It's much easier to identify a geometry than it is to identify a triangle within that geometry. From your question, I'm guessing that your hitboxes are related to FPS functionality. In that case, I think it would be a lot easier for your system to keep a head hitbox, arm hitboxes, torso hitboxes, etc. Then to determine a hit, all you have to do is cast a ray through your scene and find the first hitbox that the ray collided with. If it hit a head hitbox, it's a headshot. The trouble with identifying triangles, in addition to what @pspeed said, is that you have to do a whole lot of "book keeping" to determine which triangle belonged to which hitbox. You'll probably have to do a lot more work to correctly identify targets using your triangle system than just attaching the hitboxes to the characters. If you need more information than one hitbox per body part gives (i.e., did the shot come from the front or from behind), then use multiple hitboxes per "target" - i.e., head front, head back, torso front, torso back. If you're trying to do this to save memory, my guess is that you'll use more memory and processor time keeping track of triangles and identifying which one got hit than the memory and processor time involved in using the much simpler hitbox system I described above.