On-chain FX & multi-currency corridors

The first wave of stablecoins was dollar-only. The next is multi-currency. Circle’s Arc ships with Circle StableFX, an FX engine for USDC, EURC, and regional partner stablecoins. Quotes are priced offchain through RFQ. Trades settle onchain through Circle’s FxEscrow contract on Arc. Onchain settlement gives a treasury something it has never had: every trade’s funding and payout, for every currency pair, as a queryable event stream. This guide turns that stream into FX flow analytics, a published rate oracle, and exposure controls.

What you’ll build

Step 1: Index settlement activity

FxEscrow’s events record each trade’s lifecycle: TradeRecorded, TakerFunded, MakerDelegateFunded, and TradeStatusChanged. They carry the trade ID and the parties, but not the currencies or amounts. The money moves as ordinary ERC-20 Transfer events into and out of FxEscrow. The pipeline below captures both and writes them to two ClickHouse tables that you join on transaction_hash. FxEscrow addresses (from Arc’s contract addresses):

Network	FxEscrow
Arc mainnet	`0xe2E5F173576B513d994073CCbDaCBE027d43DFe6`
Arc testnet	`0xd68256f4D69C6BbEcB873D8588AE0Dc6B8E22E10`

stablefx-settlements.yaml

name: stablefx-settlements
resource_size: s

sources:
  fx_escrow_logs:
    type: dataset
    dataset_name: arc_testnet.raw_logs
    version: 1.0.0
    start_at: latest
    filter: address = '0xd68256f4d69c6bbecb873d8588ae0dc6b8e22e10'

  # ERC-20 Transfer events where FxEscrow is the sender or the receiver
  escrow_transfer_logs:
    type: dataset
    dataset_name: arc_testnet.raw_logs
    version: 1.0.0
    start_at: latest
    filter: topics LIKE '0xddf252ad1be2c89b69c2b068fc378daa952ba7f163c4a11628f55a4df523b3ef%' AND topics LIKE '%d68256f4d69c6bbecb873d8588ae0dc6b8e22e10%'

transforms:
  fx_decoded:
    type: sql
    primary_key: id
    sql: |
      SELECT
        id,
        _gs_log_decode(
          '[{"type":"event","anonymous":false,"name":"TradeRecorded","inputs":[{"name":"id","type":"uint256","indexed":true},{"name":"quoteId","type":"bytes32","indexed":true}]},{"type":"event","anonymous":false,"name":"TakerFunded","inputs":[{"name":"id","type":"uint256","indexed":true},{"name":"taker","type":"address","indexed":true}]},{"type":"event","anonymous":false,"name":"MakerDelegateFunded","inputs":[{"name":"id","type":"uint256","indexed":true},{"name":"makerDelegatedFunder","type":"address","indexed":true},{"name":"makerDelegatedRecipient","type":"address","indexed":true}]},{"type":"event","anonymous":false,"name":"TradeStatusChanged","inputs":[{"name":"id","type":"uint256","indexed":true},{"name":"actor","type":"address","indexed":true},{"name":"newStatus","type":"uint8","indexed":false}]}]',
          topics,
          data
        ) AS decoded,
        block_timestamp,
        transaction_hash,
        _gs_op
      FROM fx_escrow_logs

  fx_trade_events:
    type: sql
    primary_key: id
    sql: |
      SELECT
        id,
        decoded.event_signature AS event,
        decoded.event_params[1] AS trade_id,
        decoded.event_params[2] AS party,        -- taker, funder, actor, or quoteId
        decoded.event_params[3] AS detail,       -- delegated recipient or new status
        to_timestamp(block_timestamp) AS at,
        transaction_hash,
        _gs_op
      FROM fx_decoded
      WHERE decoded IS NOT NULL

  transfer_decoded:
    type: sql
    primary_key: id
    sql: |
      SELECT
        id,
        address AS token,
        _gs_log_decode(
          '[{"type":"event","anonymous":false,"name":"Transfer","inputs":[{"name":"from","type":"address","indexed":true},{"name":"to","type":"address","indexed":true},{"name":"value","type":"uint256","indexed":false}]}]',
          topics,
          data
        ) AS decoded,
        block_timestamp,
        transaction_hash,
        _gs_op
      FROM escrow_transfer_logs
      -- Arc mirrors every USDC transfer as an 18-decimal native-balance event from this address; skip it to avoid double counting
      WHERE address <> '0xfffffffffffffffffffffffffffffffffffffffe'
        AND SPLIT_INDEX(topics, ',', 3) IS NULL  -- ERC-20 only; ERC-721 Transfer has a 4th topic

  fx_escrow_transfers:
    type: sql
    primary_key: id
    sql: |
      SELECT
        id,
        token,
        decoded.event_params[1] AS from_address,
        decoded.event_params[2] AS to_address,
        CAST(decoded.event_params[3] AS DECIMAL(38, 0)) AS amount,  -- raw token units
        to_timestamp(block_timestamp) AS at,
        transaction_hash,
        _gs_op
      FROM transfer_decoded
      WHERE decoded IS NOT NULL

sinks:
  trade_events_out:
    type: clickhouse
    from: fx_trade_events
    table: fx_trade_events
    secret_name: MY_CLICKHOUSE
    primary_key: id
  escrow_transfers_out:
    type: clickhouse
    from: fx_escrow_transfers
    table: fx_escrow_transfers
    secret_name: MY_CLICKHOUSE
    primary_key: id

For mainnet, change arc_testnet to arc_mainnet and use the mainnet FxEscrow address, in lowercase, in both filters.

The ABI covers the four events FxEscrow emits during a trade. The decoder drops any event that isn’t in it, so add new events to the ABI if Circle adds them. Amounts are raw token units. Divide by the token’s decimals(): USDC and EURC use 6.

Step 2: Analyze FX flows

With trade events and token movements in your warehouse, flows per currency are SQL:

-- Funding into and payouts out of FxEscrow per token, last hour
SELECT
  token,
  sumIf(amount, to_address   = '0xd68256f4d69c6bbecb873d8588ae0dc6b8e22e10') AS funded_in,
  sumIf(amount, from_address = '0xd68256f4d69c6bbecb873d8588ae0dc6b8e22e10') AS paid_out,
  count()                                                                    AS transfers
FROM fx_escrow_transfers
WHERE at > now() - INTERVAL 1 HOUR
GROUP BY token;

To tie a token movement to a trade, join on the transaction hash. Funding transfers share a transaction with the TakerFunded or MakerDelegateFunded event for that trade:

SELECT e.trade_id, e.event, t.token, t.from_address, t.amount
FROM fx_escrow_transfers AS t
JOIN fx_trade_events AS e USING (transaction_hash)
WHERE e.event IN ('TakerFunded', 'MakerDelegateFunded');

From the same tables you get corridor flows (net USDC→EURC movement), counterparty league tables, and the funding legs of each trade, grouped by trade_id. Maintain live per-token totals with the PostgreSQL aggregate sink.

Step 3: Publish an FX rate oracle

Other contracts and apps need an FX rate. A scheduled Compose task reads the rate your desk computes from these flows and publishes it onchain, the same mechanism as the NAV oracle, pointed at an FX pair.

src/tasks/publish-fx-rate.ts

import type { TaskContext, Chain } from "compose";

// Publish to the same network your StableFX pipeline reads from (Arc). Arc
// isn't a built-in Edge RPC chain yet, so define it as a custom chain; swap for
// evm.chains.<name> once it's available on Edge. Note Arc pays gas in USDC.
const ARC_CHAIN: Chain = {
  id: 5042002, // Arc testnet
  name: "Arc Testnet",
  testnet: true,
  nativeCurrency: { name: "USD Coin", symbol: "USDC", decimals: 6 },
  rpcUrls: {
    default: { http: ["https://rpc.testnet.arc.io"] },
    public: { http: ["https://rpc.testnet.arc.io"] },
  },
  blockExplorers: { default: { name: "Arc Explorer", url: "https://explorer.testnet.arc.io" } },
};

export async function main(context: TaskContext) {
  const { fetch, evm } = context;

  const rate = await fetch<{ pair: string; rate: number; asOf: string }>(
    "https://fx.example.com/rate?pair=USDC-EURC",
  );

  const wallet = await evm.wallet({ name: "fx-oracle", sponsorGas: true });
  await wallet.writeContract(
    ARC_CHAIN,
    "0xFX_ORACLE",
    "updateRate(bytes32,uint256,uint64)",
    [
      "0x555344432d45555243", // "USDC-EURC"
      BigInt(Math.round(rate.rate * 1e18)),
      BigInt(Math.floor(new Date(rate.asOf).getTime() / 1000)),
    ],
  );

  return { pair: rate.pair, rate: rate.rate };
}

Step 4: Monitor exposure and enforce limits

FX desks live and die by position limits. Because you have every funding and payout, net exposure per currency is a running sum, and a Compose task can watch it and act (alert, hedge, or halt quoting) when a limit is breached. Compose’s durable execution guarantees the control fires, and every decision is traced for the desk’s risk review. This is a live version of the circuit-breaker pattern.

Business outcomes

Every settlement is data: funding and payout flows per currency and corridor, not a dealer’s end-of-day summary.
Multi-currency treasury: real-time positions across USD, EUR, and regional stablecoins in one warehouse.
A rate you can publish and defend: an onchain FX oracle sourced from actual settlements, with an audit trail.
Automated limit enforcement: exposure breaches trigger controls in seconds.

Resources

Can’t find what you’re looking for? Reach out to us at support@goldsky.com for help.

Get started

Solutions

Account & billing

Security & operations

API reference

Resources

On-chain FX & multi-currency corridors

What you’ll build

Step 1: Index settlement activity

Step 2: Analyze FX flows

Step 3: Publish an FX rate oracle

Step 4: Monitor exposure and enforce limits

Business outcomes

Resources

​What you’ll build

​Step 1: Index settlement activity

​Step 2: Analyze FX flows

​Step 3: Publish an FX rate oracle

​Step 4: Monitor exposure and enforce limits

​Business outcomes

​Resources

What you’ll build

Step 1: Index settlement activity

Step 2: Analyze FX flows

Step 3: Publish an FX rate oracle

Step 4: Monitor exposure and enforce limits

Business outcomes

Resources