> ## Documentation Index
> Fetch the complete documentation index at: https://docs.cryptocheckout.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Open decisions

> Commercial and design choices that are genuinely undecided, with the trade-offs and what each blocks.

These are not oversights. Each is a real choice with defensible answers on both sides.

## Who covers the settlement cost

**The question.** Does the customer's quoted amount include the gas, or does the merchant absorb it?

The contract mechanism is **identical either way** — it is a quoting-policy flag, not a redeploy. Build the mechanism, decide the incidence later.

| Option           | \$20 TRON order                            |
| ---------------- | ------------------------------------------ |
| Merchant absorbs | Customer pays $20.00, merchant nets $17.96 |
| Customer covers  | Customer pays $21.84, merchant nets $19.80 |

Customer-covered is symmetric with the connect rail, where the customer already pays their own gas, and it uses machinery already being built. The cost is a visible 9% surcharge on a small TRON order, which will hurt conversion — though that is TRON telling the truth about itself.

## Fee disclosure

**The question.** How is a chain-dependent cost published?

Actual-cost billing makes the merchant's effective rate vary by chain: 1% plus $0.01 on Ethereum, 1% plus $1.84 on TRON. Treasury also starts receiving "0.75% plus documented settlement costs" rather than a flat 0.75%.

This collides with a single all-in headline number. The fee disclosure has already needed one correction — the site advertised 99% when merchants actually keep 98.505% — and `feeDisclosure.test.ts` now asserts the published copy against a re-implementation of the split.

A published **flat per-chain surcharge** may sell better than a true pass-through even though it over- and under-recovers at the margins.

## The distributor reward

**The question.** Keep the 50 bps, cut it, or re-frame it?

Keeping it means the honest headline is 1.495%, not 1%. Cutting it restores a clean 1% flat but removes the incentive for anyone to call `distribute()` — and, until the reimbursement mechanism lands, removes the platform's only gas-recovery channel.

## Native-gas exposure

**The question.** How large a buffer, and does it get hedged?

The keeper spends ETH and TRX and recovers USDC. Even with a perfect USD snapshot at spend time, the platform is whole in dollars while its gas balance depletes and must be refilled at a later price.

<Warning>
  This is a **permanent short position in ETH and TRX**, linear in volume. Roughly $1,840 per month of float at 1,000 TRON payments; a 30% TRX move costs about $550. It does not shrink with better engineering — every other cost in this section does.
</Warning>

The recommendation is to treat native gas as **inventory**: hold a buffer sized to expected monthly spend, pre-buying the exposure, rather than hedging it as a trade.

This couples directly to the distribute cadence. Retiring the eager five-minute cron for a monthly backstop stretches the float from about five minutes to 31 days. **The cadence decision and the treasury exposure are one decision** and were originally written as two independent issues.

## Quote-to-sweep gas drift

**The question.** Who absorbs a gas move between quoting and sweeping, and how much headroom?

Ethereum moved from 0.062 to 0.147 gwei inside 40 minutes on 2026-07-30. The merchant's tolerance is enforced at quote; the overflow lands on treasury within the immutable cap. A roughly 1.5× multiplier on Ethereum only is recommended; L2 and TRON drift are immaterial.

## Refund fees

**The question.** Does a refunded order get its platform fee back?

Refunds require distribute and claim to have run first, so the fee has already been taken. Whether it is returned, and how that is disclosed, is undecided. See [Finality and refunds](/settlement/finality-and-refunds).

## TRON claim cost

**The question.** Is \$4.59 per claim acceptable, and can it be reduced?

The merchant signs from their own wallet and therefore burns at 100 sun. It amortises over the batch — about \$0.23 at 20 orders — which makes claim cadence a real merchant-facing economic decision.

Options exist: delegate rented energy to the merchant for the claim (but then the platform pays), or batch harder. Neither is designed.

## Where the abuse gate should sit

**The question.** Is TRON's \$27 pool deploy a wall to demolish or a filter to keep?

Currently resolved as a **filter to keep** — it is the anti-abuse gate on the chain where abuse costs most. That resolution is what rejected the TRON pool clone. It should be revisited if merchant acquisition data shows the wall is actually costing signups, but note that the merchant-paid gas design already removes most of the abuse exposure it was built for.

## Blocked on measurement, not judgement

These are not open decisions — they are unanswered facts. See [Status](/overview/status).

* Dynamic Energy Model attribution: implementation or proxy
* Like-for-like forwarder sweep gas, measured on one harness
* `0x41` CREATE2 re-derivation for any new forwarder init code