I'm a bit puzzled. Suppose you do single Kyber key exchange, and you then hash the shared secret with a domain separator to get a fingerprint, wouldn't that mea

I'm a bit puzzled. Suppose you do single Kyber key exchange, and you then hash the shared secret with a domain separator to get a fingerprint, wouldn't that mean you now have a shared secret that you can verify wasn't under MITM. You then can build post quantum future secrecy / key rotation on top of that, by mixing in new key material and it remains secure from MITM, as long as the internal state of the endpoint isn't compromised. The endpoint compromise is outside networked TCB threat model, as such compromise could also be used to exfiltrate long term post-quantum identity keys for undetectable MITM).