This really applies to more than just code. Any time you rely on an outside provider you're opening up yourself to risks. Ultimately it comes down to whether or not you trust the ability of the provider to provide you with service more than you trust your own ability.

Really though, this situation is just re-hashing the fact that everyone has to be careful about single points of failure. If you're screwed because a single service provider you rely on is down you need to think about increasing redundancy. This goes for absolutely everything whether the service is for DNS, VPSes, or version control repositories.