I actually thought about adding wiring to elements. That's not so much of a difficulty to implement. Then one could play a bit with the places cables pass...
The joints interference you can actually see them, as you can control each joint with sliders in the Interaction » Controllers menu. The quality of the simulation has some problems still.
I agree that robot simulators can be precious to help implementing control algorithms, try several possibilities and scenarios, teach the robot, as in this experiment https://assemblino.com/assemblino.html?id=20123372
(requires WebGL and Chrome). The model is rough, and the neural network is trained with a relatively small set of examples, but the point is just to show, give an example, of how a simulator can be used for training.
The online tool, in the website is useful mainly to share components and show ideas, access them anywhere, let others tweak and play with it. Nonetheless you can use your own IDE to program components and make assembles locally and interface assemblino in a local port with arduino boards. This is documented.
As this is an open source, free software (MIT license, do whatever with it), you can follow the development on GitHub. Meanwhile I will make improvements on some aspects during the following months. At this stage what you see is just a bit more than the backbone, which actually gave a lot of work. Most of the work further on is to program components, like sensors and actuators, maybe some already built assembles, and some controllers or intelligent algorithms.
Thanks for your feedback. I appreciate it very much! I will try to blog a bit around this.