Patcher currently uses the mail buffers as “master” buffers for controlling all operations: building a patch, creating the ChangeLog entries, committing... all is done from the mail buffer. Note however that you don't need to actually send mails to use Patcher (see Fake Mail Method).
To use Patcher on a certain project, you start by preparing a (possibly fake) mail. There are several ways to do so: you could start a brand new message, “adapt” a message already in preparation to Patcher, or even compose some sort of a Patcher reply to another message.
At any time from a mail buffer, you may change your mind and decide that
starting Patcher was a mistake. You can then call the function
patcher-mail-kill, bound to C-c C-p k, and Patcher will
“kill” the current project, cleaning up the place like Patcher had
never existed before.