Does a Merkle docket attest the transaction, or the workflow decision that produced it?

[source-origin]

The blueprint + docket model is the part I keep thinking about. A blueprint defines a multi-step, multi-party flow as declarative JSON, and transactions ("every transaction is signed, every change is immutable") are sealed into Merkle-tree dockets. The seam I can't resolve from the docs is what the docket proof actually attests.

A blueprint step makes a decision (conditional routing, business-rule evaluation) and that decision causes a transaction to land. Those can diverge in the usual ways: a retry, a partial step, a routing change replayed against a newer blueprint version. So an inclusion proof for transaction T in docket D proves exactly one thing — that T is in D. It does not, by itself, prove which blueprint version and which rule evaluation authorized T.

Two questions:

1. Is there a signed artifact binding the workflow decision to the transaction (so a verifier holds decision + transaction and recomputes the link), or is the signed transaction the sole bearer of authority?
2. Since blueprints are versioned and mutable, does the docket commit the blueprint/decision context under which T was sealed — so a third party can tell which definition was in force at seal time, rather than validating an old inclusion proof against the current definition?

If that binding is intentionally left to the application layer, that's a fine answer — I'm trying to pin down where the join point lives.