As you chose NOT to effectively answer the question, I can't really justify even looking into whether or not this is practical.  To be honest, this looks more like hacking research... to learn how an attack vector might be easily hidden.   I can't even come up with one single benefit of doing this that doesn't involve hacking research.   Until you can be more forthcoming, I'm not even remotely interested.