Session key through ECDH

[dholms]

I'm doing AWAKE with Elliptic curve keys as the temp keys (NIST P-256 to be precise, still from WebCrypto).

With ECDH, you can calculate a shared secret between two keypairs. Is it reasonable to use this as the session key?

The whitepaper says:

The AES256-GCM session key MUST be included in the "facts" (fct) field.

If we have a calculated shared secret as the session key, I don't think that we need to include it in the UCAN, the only thing you need to attest to is the tempDID from the provider.

My instinct is to send as the session negotiation a message that includes the providers tempDID as well as the specified UCAN with no att & the providers tempDID as the audience. This seems to match the same authn/authz guarantees as the specified protocol over RSA. Thoughts?

If this seems reasonable, this could be good food for thought for a more fleshed out spec that allows for different key sharing algos.