Files encrypted by crypt are compatible with those treated by the editors ed(1), edit(1), ex(1), and vi(1) in encryption mode.
The security of encrypted files depends on three factors: the fundamental method must be hard to solve; direct search of the key space must be infeasible; ``sneak paths'' by which keys or clear text can become visible must be minimized.
crypt implements a one-rotor machine designed along the lines of the German Enigma, but with a 256-element rotor. Methods of attack on such machines are known, but not widely; moreover the amount of work required is likely to be large.
The transformation of a key into the internal settings of the machine is deliberately designed to be expensive, that is, to take a substantial fraction of a second to compute. However, if keys are restricted to (say) three lower-case letters, then encrypted files can be read by expending only a substantial fraction of five minutes of machine time.
If the key is an argument to the crypt command, it is potentially visible to users executing ps(1) or a derivative. The choice of keys and key security are the most vulnerable aspect of crypt.
If output is piped to nroff and the encryption key is not given on the command line then do not pipe crypt through pg(1) or any other program that changes the tty settings. Doing so may cause crypt to leave terminal modes in a strange state (see stty(1)).