[Agentproto] Introduction, and a scope question on vocabulary semantics

[Cruce Saunders <[e-mail]>]

Hi all,

New here. I subscribed after finding the BoF materials, and I'm glad to
have found this before the charter settled.

Brief context: I work on schema and vocabulary alignment between systems,
at ARAMAI. I've read the draft charter, Brian's simplification PR, and the
discussion.

So, I work on schema interoperability between enterprise agents.

A scenario that keeps coming up: two agents transact. Discovery works, the
handshake is clean, both send "price." One means with tax, one means
without. Every check passes — the schema confirms the shape, attestation
confirms the identities. The transaction succeeds and the books are wrong.

Ted's frame-of-reference marker is the nearest thing in the thread.
Attestation can prove a marker genuinely belongs to a frame. But nothing in
the current scope checks that both parties read the fields inside it the
same way.

Is that inside or outside this group's scope? Out-of-scope may well be the
right answer for a tight charter. But if it's out, it's worth one sentence
saying so, thus implementers know checking shared meaning is their job, not
the protocol's.

I've filed an issue on the charter repo with a proposed sentence, so
there's concrete text to accept or reject - issue #80
The out-of-scope list contains "agent behavior, decision-making, or
planning semantics." Understand and agree. But whether two agents interpret
the same field the same way seems to stand apart. e.g. Two agents complete
a correct handshake and exchange a field named "price." One includes tax,
the other doesn't. Structure validation passes, the exchange succeeds, the
result is wrong.

So: in scope, out of scope, or somewhere else entirely?

Thanks,
Cruce Saunders
ARAMAI

ps I wrote up some more on this kind of agreement making here in a
preprint  - https://zenodo.org/records/20754007