transactions – Is that this dummy-padded Ordinals atomic-swap PSBT building appropriate and secure?

I’ve constructed an Ordinals atomic-swap market. The core (settlement.js,
~260 traces) assembles a single unsigned transaction, co-signed by each events:
inputs [buyer dummy, seller inscription, buyer payment]; outputs
[inscription→buyer (dummy+inscription value), payment→seller, change].

A sat-flow verifier asserts the inscribed sat lands within the purchaser output earlier than
signing. It is confirmed end-to-end on regtest with an actual ord inscription; it
runs on signet/regtest solely and mainnet is code-locked.

Repo: https://github.com/Marvintymo/asset-loop

Particular questions:

1. Is the dummy-padded sat routing appropriate for non-zero inscription offsets and
   multi-inscription UTXOs?

2. I signal the fully-assembled tx with SIGHASH_ALL (no pre-signed itemizing). Is the
   vendor absolutely protected vs. the OpenOrdex SINGLE|ANYONECANPAY mannequin?

3. For taproot inputs added witnessUtxo-only, what faucet fields have to be current
   for arbitrary exterior wallets (Unisat/OKX/Xverse) to finalize reliably?

4. Is my payment mannequin (68 vB/in, 43 vB/out, +11, ×1.1) secure throughout enter sorts?

5. Finest observe for affirmation depth + double-spend checks earlier than broadcast?

Thanks upfront!!