nick-doc/08 - Operations/Handoff - RN Multichain Probe - 2026-05-28.md

Handoff: RN Multichain Proxy Probe — 2026-05-28

Probe execution

Script: backend/scripts/probe-rn-chains.ts (commits 01b9ea0 → 4a85737, backend 2.6.46 → 2.6.47) Executed: 2026-05-28T15:57:33Z (final run after Base fix)

Results summary

Chain	Chain ID	Proxy Address	RPC	Reachable	Has Code	Call Valid	Status
BNB Smart Chain	56	0x0DfbEe143b42B41eFC5A6F87bFD1fFC78c2f0aC9	publicnode	✅	✅	✅	Verified
Arbitrum One	42161	0x0DfbEe143b42B41eFC5A6F87bFD1fFC78c2f0aC9	publicnode	✅	✅	✅	Verified
Ethereum Mainnet	1	0x370DE27fdb7D1Ff1e1BaA7D11c5820a324Cf623C	publicnode	✅	✅	✅	Verified
Polygon	137	0x0DfbEe143b42B41eFC5A6F87bFD1fFC78c2f0aC9	publicnode	✅	✅	✅	Verified
Base	8453	0x1892196E80C4c17ea5100Da765Ab48c1fE2Fb814	publicnode	✅	✅	✅	Verified

Methodology

1. Reachability: eth_blockNumber against each chain's public RPC endpoint.
2. Code presence: eth_getCode at the candidate proxy address.
3. Function validity: eth_call with a dummy transferFromWithReferenceAndFee payload. A valid proxy reverts with an ERC-20/execution error (proving the selector is recognized). An invalid address would revert with "unknown function selector" or return empty.

Key findings

• BSC / Arbitrum / Polygon: Canonical CREATE2 proxy 0x0DfbEe143b42B41eFC5A6F87bFD1fFC78c2f0aC9 is deployed and responds correctly. ✅
• Ethereum Mainnet: v0.1.0 proxy 0x370DE27fdb7D1Ff1e1BaA7D11c5820a324Cf623C is the official deployment per RN smart-contracts artifact. The canonical CREATE2 address is also deployed (v0.2.0-next) but the artifact marks it as an additional deployment. ✅
• Base: The canonical CREATE2 address has no code on Base. However, RN does support Base via a non-canonical deployment at 0x1892196E80C4c17ea5100Da765Ab48c1fE2Fb814 (per RN smart-contracts artifact v0.2.0, packages/smart-contracts/src/lib/artifacts/ERC20FeeProxy/index.ts). ✅

Base proxy address hunt

Source: https://raw.githubusercontent.com/RequestNetwork/requestNetwork/master/packages/smart-contracts/src/lib/artifacts/ERC20FeeProxy/index.ts

base: {
  address: '0x1892196E80C4c17ea5100Da765Ab48c1fE2Fb814',
  creationBlockNumber: 10827274,
},

Action taken

• Updated supportedChains.json with correct per-chain proxy addresses.
• Restored Base USDC/USDT entries to tokens.json.
• All 5 chains now active in the registry.

AML scope note (for legal / compliance review)

The current AML implementation (Task #10, shipped in backend/frontend 2.6.47) performs sanctions-only screening via the Chainalysis Public Sanctions API. It checks whether a buyer's source wallet address appears on known sanctions lists (OFAC, UN, HMT, etc.). It does not perform full AML risk scoring — there is no transaction clustering, entity attribution, travel-rule monitoring, or behavioral risk scoring. Upgrading to comprehensive AML/KYT would require a paid Chainalysis KYT tier (or equivalent provider such as Elliptic, TRM Labs, or ComplyAdvantage), which runs ~$100K+/year for production volumes and requires an enterprise contract. The sanctions-only tier is free (5,000 requests per 5 minutes) and is the correct scope for a v1 compliance posture, but it should be explicitly described to regulators/customers as "sanctions screening" rather than "AML screening."

Remaining work

• BSC USDT paid end-to-end probe (PRD §2 AC #3) — pending human-in-the-loop.
• Mainnet USDT approve(0) reset verification (PRD §2 AC #4) — VERIFIED via anvil fork test.

Human-blocked items (requires owner with wallet on dev)

These three items cannot be validated by automated tests alone. A human with a funded wallet on the dev environment must execute each probe before the corresponding feature is considered production-ready.

#	Item	Precise next step	Blocking
1	Task #7C — Live multi-seller divergent-destination probe	Create a cart with seller-offers from ≥2 different sellers, complete checkout, verify RN creates 2 separate Payments with 2 distinct derived destination addresses, and both webhooks fire correctly.	Task #7 closure
2	Task #8 — BSC USDT paid end-to-end probe	On dev.amn.gg, complete a real BSC USDT pay-in through the in-house checkout (approve + transferFromWithReferenceAndFee), confirm webhook marks Payment completed, and BscScan shows the token transfer.	Multichain release gate
3	Task #11 — Trezor signing dry-run	Register a physical Trezor via /api/trezor/register, build a sweep tx via POST /api/admin/actions/build-tx, sign it on-device through the admin UI, broadcast via wagmi, and confirm POST /api/admin/actions/confirm-tx accepts the Trezor proof.	Trezor enforcement toggle

Mainnet USDT approve(0) reset — fork test verification

Test: scripts/tenderly-usdt-reset-test.sh (anvil fork of Ethereum mainnet) Date: 2026-05-28

Setup

• Forked mainnet at https://ethereum-rpc.publicnode.com
• Impersonated whale 0xF977814e90dA44bFA03b6295A0616a897441aceC
• Transferred 100 USDT to test buyer 0x0000…dEaD

Transaction sequence

Step	Description	Tx Hash
1	Transfer USDT whale → buyer	0x574440bc7aa2915ff8b5adddc9b083420c5e426894fe98d7c72196f7d8e37c22
2	approve(proxy, 50 USDT) — quirk setup	0x4dfef21e19e9fe17e7fb60ecefdce6f5a240e74de904f772c57ad3a3013454b8
3	approve(proxy, 0) — reset	0xf249a05db18753abf7625b44074b81939764c5a1590928d29aba0757ae5446b0
4	approve(proxy, 100 USDT) — re-approve	0xd27c1c382a20db754dd3d67efbb565016ad33c460be9fb7a09986589626aaef5
5	transferFromWithReferenceAndFee — pay	0xf7b44000fc11bcc1e832360acc9cbbfccd8be4c21f5940df00115ea7f2c4038a

Results

• Allowance after partial approve: 50,000,000 ✅
• Allowance after reset: 0 ✅
• Allowance after full approve: 100,000,000 ✅
• Buyer balance decreased by exactly 100 USDT after payment ✅

Verdict

PRD §2 AC #4 status: VERIFIED — The USDT-mainnet approve(0) reset flow executes correctly on a mainnet fork. All five transactions succeeded. The two-step approve pattern (reset → re-approve → pay) is validated end-to-end.