The exchange now tells you what was actually checked

We spent this stretch on the native-asset exchange: a labelled activity feed, a global transaction section, working failure messages, and Sentinel screening on withdrawals. The most interesting part was deciding what not to claim.

A tick you cannot verify is worse than no tick

It would have been easy to put a small green shield on every exchange operation. It looks reassuring, it costs nothing, and nobody would question it.

The problem is that most exchange operations are not screened. Sentinel scores ordinary transfers and NRO withdrawals. Creating a pool, swapping, adding liquidity — none of these pass through it today, and pretending otherwise would be a claim the user has no way to check.

So each transaction now carries its real state: whether it is eligible for screening at all, whether screening is enabled, and whether it was actually held with a score. Where nothing was checked, the interface says so.

Why those operations aren't screened yet

Not an oversight. Sentinel's features are built around a wallet: balance, nonce, the share of a balance being moved, spending velocity. For a swap of one asset into another, those inputs either do not exist or mean something different. A number produced from them would look like a risk score without being one.

Screening exchange operations properly needs its own features — size against pool reserves rather than against a wallet, price deviation, how fresh an asset is — and its own calibration. That is real work, and it is queued as such rather than faked.