Unless the data is random (which is to say the additional layer of crypto is probably not buying you much) you can do e.g. frequency analysis on the output of t

Unless the data is random (which is to say the additional layer of crypto is probably not buying you much) you can do e.g. frequency analysis on the output of the cipher. If you have a reasonable sized corpus (given how biased the PRNG is, probably a few thousand characters is overkill) you can probably break it effortlessly.