I assume you meant to say that the private key cannot feasibly be derived from the public key, but the other way around is trivial.
Nice try, General Alexander. :P
In DSA, the public key is (p,q,g,y) and the private key is (p,q,g,x). The public "y" is computed directly from the private key using the formula: y = g^x mod p. Thus, anyone who has the private key can compute the public key.
In RSA, the public key is (n,e) and the private key is (n,d). They're not interchangeable because of two speedups that we use:
1. Small public exponents. Most of the time, e is either 65537, 5, or 3. This speeds up encryption and signature verification, but also means you can trivially guess the public key.
2. CRT exponentiation. The private key is actually (n,d,p), where p is one of the prime factors of n. This speeds up decryption and signature verification, but recall that the security of RSA is based upon the difficulty of factoring n.
ssh-keygen -y -f private_key.pem > public_key.pubThe public key file doesn't.
It wouldn't need to, though. Given the private key (n,d), the public key is probably (n,65537).
Still, if someone puts your pub key in an authorized_keys of a malicious host, and then manipulates your DNS or ARP to conduct you to such host, you should get some SSH warnings about the host fingerprint, etc.
And still, if you're using a dedicated key, the only possible attack could be to make you commit your repo changes, to the malicious machine instead of to github.
At least when you use agent forwarding, your key is only exposed while you're actually connected to the machine.
I'm a bit concerned about the public keys public url. Ok, it's just public keys, so it must be safe, but I wonder : is it safe like it's "safe" publishing a login without a password ? In other word, doesn't knowing public key make easier to break private one ?