Why this needs buildingMeasured on chain
Everything above is a proposal. This section is not. It is a scan of
Base over
days, reading every
StrategyReported event and attributing each to the account
that paid its gas. Coverage
, no API key.
| Keeper account | Vaults | Reports | Share of all reports |
|---|
What the scan found
of vaults were reported by exactly one address for the entire window — not one had a second key step in. The median gap between reports is ; the longest any vault went unattended was . Nothing reverted and nothing alerted. It simply stopped compounding.
What it does not show
It is not proof any vault is unsafe — a well-built one survives its keeper stopping, and the cost is foregone yield rather than principal. It counts one event on one chain, so the vault figure is a floor. And a key is not an operator: addresses link to a block explorer and are never attributed to a company.
Full method, the cadence distribution, the silence ranking and a panel that re-runs this query live in your browser are on the sister site, NightShift.
Name the keeper. Bond it. Slash it.
Nothing here is built yet. The problem it answers is on chain today and you can check it yourself.