# Before you list a MiniPay Mini App: what I'd check in the contracts and the backend

**URL:** <https://forum.celo.org/t/before-you-list-a-minipay-mini-app-what-id-check-in-the-contracts-and-the-backend/13988>\
**Category:** Developers\
**Created:** [November 5, 2026, 8:43am UTC](https://forum.celo.org/t/before-you-list-a-minipay-mini-app-what-id-check-in-the-contracts-and-the-backend/13988 "2026-11-05T08:43:57Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![Zealynx](https://sea2.discourse-cdn.com/flex020/user_avatar/forum.celo.org/zealynx/32/12323_2.png) [@Zealynx](https://forum.celo.org/u/Zealynx)\
**Post date:** [November 5, 2026, 8:43am UTC](https://forum.celo.org/t/before-you-list-a-minipay-mini-app-what-id-check-in-the-contracts-and-the-backend/13988/1 "2026-11-05T08:43:57Z")

</div>

[MiniPay’s Q3 update](https://forum.celo.org/t/minipay-update-q3-2026/13926) says 77 Mini Apps are live, and a lot of the people building here are working on the next ones. I went through the MiniPay developer docs and the listing requirements the way I’d read a codebase before an audit. The docs are good on UX. Security is where a Mini App differs from a normal dapp, and that difference is easy to miss when everything works fine in testing.

### Your backend doesn’t know who is calling

MiniPay connects on page load, and the [best practices page](https://docs.minipay.xyz/getting-started/best-practices.html) says not to ask users to sign a message to access the app or to authenticate. Nice for users. But if your app has a backend, the wallet address it gets is just whatever the request says. Anyone with curl can send someone else’s address.

For reads that’s usually fine, addresses are public. It’s a problem when the request moves something of value: rewards, points, cashback, a prize claim, a change of payout address, private order data. Those belong in a contract call, where `msg.sender` does the work, or your backend should act only after it reads the user’s own transaction from the chain.

`window.ethereum.isMiniPay` has the same limit. It’s a property on an object in the browser, so it tells you the environment and nothing else. A script on a laptop can set it, so a reward shouldn’t depend on it.

### One price, tokens with different decimals

The [docs](https://docs.minipay.xyz/technical-references/retrieve-balance.html) ask you to use USDT, USDC and USDm and, ideally, adapt to whichever one the user holds most of. USDC and USDT have 6 decimals and USDm has 18. So if a contract stores one price and takes the token as a parameter, something breaks. Store 5 dollars as `5e6` and a USDm payer sends 0.000000000005 USDm for the same item. Store it as `5e18` and the USDC payer just reverts.

Keep a price per token, or normalise with `decimals()` over a fixed allowlist, and run your tests with all three, not only the one you happen to have on testnet.

### The fee can come out of the token being spent

MiniPay uses your `feeCurrency` only if the wallet has a [non-dust balance of it](https://docs.minipay.xyz/technical-references/send-transaction.html). If not, it pays with the USD stablecoin the wallet holds most of. On Celo the maximum gas cost is [reserved with `debitGasFees`](https://docs.celo.org/build/fee-abstraction/overview) before the transaction runs, and the unused part comes back after. So a “pay all” button that spends a user’s full USDT balance can fail when USDT is also paying the fee, and the user just sees a failed payment. Leave room for the fee in that flow. A frontend check of `balance >= amount` doesn’t cover that case either.

### The 24 hour rule

The [listing rules](https://docs.minipay.xyz/getting-started/submit-your-miniapp.html) say you have to fix reported critical issues within 24 hours, or the listing may be temporarily disabled. Your contracts must also be verified on Celoscan, with a sample transaction for every user-facing method.

Think about those 24 hours before you deploy, because your design decides what a fix can be. An immutable contract with no pause can’t be fixed in a day. You can point the frontend away from it, but the funds stay inside. A proxy behind a 48 hour timelock doesn’t fit either. I’d want a pause role with a narrow job, held by signers you can actually reach on a Sunday, and a written plan for the funds that are inside when you pause.

### Your frontend sits inside the payment flow

There’s no address bar and no connect step, and the scripts on your page build the transaction the user confirms. A compromised npm package or third-party script is right there next to the pay button. The same submission page already asks for exact versions, a minimum release age of 7 days, `ignore-scripts=true`, a committed lockfile and a manifest of every origin the app loads. I’d treat that section as security work, and look hard at each analytics or chat widget that ends up on that manifest.

If you’ve shipped a Mini App with a backend: how does it know which wallet is calling, given there’s no sign-in?
