Crypto Algorithm Evaluation - Requesting Evaluation of Randomized Encryption Format

I have designed, over the past year, a randomized encryption format. I call it a format and not an algorithm because it's a container in which existing algorithms are plugged into. As I am travelling right now, I do not have the C# source code on hand.

The basic summary is that encryption and HMAC keys (and IVs) are randomly generated, concatenated, and encrypted into a header before the ciphertext. The plaintext is then encrypted with the randomly generated key/IV. HMAC support is included.

Is this container format already broken, or is it worth further review?

The container isn't broken, this is how lots of public key crypto is performed, slow asym encryption of a random sym key with the data then encrypted with the quicker sym algo

This is a fairly common scheme for storing the extra HMAC and IV information along with the file. The scheme sounds common, likely the weakest point is the random number generation, since it's easy to mess that up.