Hmm, I might not have read your last post correctly. I thought you meant that you only wanted to deal with the public keys, but now I'm thinking that you meant that you thought that the man page only dealt with public keys.
Here's a slightly more complete description of the process so that you know how to look at it:
-on the client:
-run ssh-keygen -t rsa
-answer the questions and it'll make ~/.ssh/id_rsa and ~/.ssh/id_rsa.pub
-id_rsa is the private key. Permissions will already be set properly but (especially if you used a blank passphrase) it must never be readable by anyone but you
-id_rsa.pub is the corresponding public key
-copy it to the server and append it to ~/.ssh/authorized_keys (on a new line). Doesn't matter who can read it, as long as only you can write it.
Now, when you ssh from the client to the server (or sftp or scp), it should just go without any prompting, unless you chose a non-blank passphrase, in which case you'll be prompted for it every time. The reason you want this is that root on the client machine can always read your private key so, if that's not you, you'll want it encrypted. To make that work with scripts, you start ssh-agent and tell it the passphrase. It then runs in the background and puts a pipe in /tmp that ssh/sftp/scp talk to to get the private key while running from within the script. So root can still grab your key via the pipe, but only so long as you have ssh-agent running.