The gate before
the payment.

No, and we will never claim to. We probed a nonsense function selector against three things on Base: USDC reverted, WETH returned empty, and a plain wallet also returned empty. So an empty answer means “no such function” on one contract, “a catch-all fallback swallowed the call” on another, and “you mistyped the address” on a third — and real functions revert on malformed calldata, so a revert proves nothing either. We can show you a token HAS a pauser, because a getter that answers with a live account is proof. Absence is not provable with the reads available, and “no dangerous privileges” is the most dangerous sentence this product could print.

Because the block-explorer API refuses free access on Base, and refuses it silently — it answers HTTP 200 with a body reading status 0, message NOTOK. A client that checks whether the response was OK reads that refusal as data. We hold no paid explorer key, so age, first and last activity, transaction count and holder concentration are absent. They are listed in every response under notChecked with that reason, rather than estimated, rather than left quietly empty.

That every signal which resolved looked ordinary — nothing more. It is a band computed over the signals we could observe, and the response tells you exactly how many that was and which ones it was not. Read notChecked before you treat clear as an all-clear. This is a diligence signal, never a verdict on a counterparty, and never financial advice.

By a fixed arithmetic rule, not a model. Score equals 100 times the sum of value times weight, divided by the sum of weights, over resolved signals only — and any hard failure, which today means an OFAC hit, forces the high band outright regardless of the arithmetic. A signal we could not observe leaves the denominator entirely; it is never scored as a zero, because that would punish an address for our blind spot. Every signal comes back with its weight and its evidence, so you can recompute the number yourself.

We can only tell you nothing turned up in the standard slots. We read the EIP-1967, EIP-1822, ZeppelinOS, beacon and Safe singleton pointers — a set, not one, because probing EIP-1967 alone reports USDC as non-upgradeable, and USDC is a proxy: its implementation lives in the older ZeppelinOS slot. A bespoke proxy that keeps its pointer somewhere else would not show up in any sweep we can do, so the evidence line says nothing was found in the standard places rather than claiming the code is immutable.

By actually doing the transfer, in a simulation. An eth_call state override lets us replace an address's code and storage for the duration of one call, so we fund a probe contract with the token, have it transfer to a second address, and measure what arrives. It was verified in both directions before shipping — USDC delivered in full, and a synthetic token built to keep ten per cent was measured at exactly ten per cent. The negative control is what makes the clean result mean anything.

Because OFAC tags digital-currency entries by ledger and publishes no BASE tag. A sanctioned party controls the same private key on every EVM chain, so screening a Base address against the ETH-tagged entries is the correct thing to do — but it is an inference, and the evidence line says so rather than implying a Base-specific list exists. We also validate the list's size and shape before trusting it, because a truncated download of a large file still returns HTTP 200.

Because the mixed case IS a checksum. EIP-55 encodes one in the capitalisation of an otherwise case-insensitive string, so a mixed-case address that fails it has had a character transposed or dropped somewhere in transit. We refuse the request free, before any price is quoted, rather than confidently reporting on whatever address that typo happens to be. An all-lowercase address asserts no checksum and is accepted as-is.

5 Ounie credits for an address and 8 for a token contract — $0.05 and $0.08 in plain money, from the same wallet as every other Ounie app. No subscription, no minimum. Ask for kind auto and we reserve the larger price but bill whichever it turned out to be, refunding the difference to the exact pool it came from. Re-reading and listing checks you have already paid for is free forever.

Yes, and that is the answer worth the most. A high band is the reason to run the check at all. What is never charged is a check where the chain would not answer — that observed nothing, so it refunds in full on the credit rail and is refused before settlement on x402, where nothing can be refunded once it has moved.

Yes. The x402 endpoint takes USDC on Base with no account and no key — and the caller is already a wallet holding USDC on Base, so it is the one endpoint in the fleet whose audience is exactly its payment method. Pass kind explicitly there, as address or token: the two cost different amounts and a pay-per-call price has to be knowable before you sign for it. Auto stays on the credit rail, which can refund.

Base mainnet, and only Base. Every reader here was verified against it — the RPC failover, the proxy slot set, the delegation designator format, the state-override simulation. A chain we have not verified would produce answers we cannot stand behind, so it is refused rather than attempted.