I never planned such tight binding (implementing interfaces across different script runtime boundaries), as it bounds to be pretty complicated, so question is, what kind of design decisions led you to such requirement?
It is again those Critter Action Behaviour Trees I am working on. I already have system that manages these actions (let's call it Manager) and set of actions in AngelScripts.
What I am thinking is reimplementing Manager in Mono, but to be able to reuse those existing actions by introducing interface in Mono, which existing actions would implement, so I could reuse those actions in Manager in Mono.
The second reason is not to be bound to Mono code, as maybe someone does not want to use it in their project and he can implement actions in AngelScript.
And the third reason is the ability to easily switch between the AngelScript Manager and Mono Manager to e.g. see the difference in performance or just to have fallback Manager in case that your Mono project won't be maintained anymore (not compatible with the newest revision of SDK as it happened after UTF change or you will discontinue the project).
I was trying something like that in C++, so I introduced interface which is then implemented in AngelScript (already existing actions), and create C++ Manager for the actions, but I am struggling with it a bit as I am not advanced user of C++ (as being Java focused), but that's another story