claimsStarted is the finality latch: once value has moved, the moderator can no longer re-flag the outcome (DESIGN §4 — "finality is value-movement"). Every value-moving entrypoint sets it — except sweepUnclaimedBonus, which deliberately does not (to stop a 1-wei donation from griefing the re-flag window).
When riskWindowStart == 0 (no active-risk window was ever observed — nobody poked during UNDER_ATTACK, or the registry reached CORRUPTED fast), sweepUnclaimedBonus sweeps the entire bonus to recoveryAddress and decrements live totalBonus to zero — a major value movement that does not latch claimsStarted. The moderator can then still make the normal §4 correction from SURVIVED to good-faith CORRUPTED (naming the whitehat who legally rescued the in-scope contract). That re-flag re-snapshots snapshotTotalBonus = totalBonus, which is now zero, so bountyEntitlement = snapshotTotalStaked — the whitehat is entitled to principal only. Per DESIGN §12 the whole pool (principal + bonus) is the whitehat's bounty; the bonus has been captured by the sponsor's recoveryAddress.
The code comment at the decrement even states it drops totalBonus "keeping the accounting honest for any later re-snapshot" — it anticipates the re-snapshot but does not account for it under-funding the whitehat.
Likelihood:
Occurs when riskWindowStart == 0 (the pool never observed the active-risk window — realistic when the whitehat operated at the agreement level and never poked this third-party pool, or the registry reached CORRUPTED quickly).
Occurs when the moderator makes the NORMAL §4 correction — an initial SURVIVED judgement (breach looked out-of-scope) later corrected to good-faith CORRUPTED (determined in-scope). This is the exact use the correction window is designed for, seconds apart — no 180-day wait.
Occurs while the sponsor (or anyone — the sweep is permissionless) calls sweepUnclaimedBonus in the gap; the sponsor is economically incentivised (they own recoveryAddress).
Impact:
The moderator-named whitehat, who performed a legal Safe-Harbor rescue, loses their entire bonus — which can be the majority of the reward the sponsor put up to attract whitehats. It is captured by the sponsor's recoveryAddress.
The stated DESIGN §4 invariant ("finality is value-movement; claimsStarted latches") is violated: the whole bonus leaves the pool without latching, and a subsequent re-flag silently under-funds the bounty.
test/audit/RVS1.t.sol — the leak plus a control proving it is specific to riskWindowStart == 0:
Run: forge test --match-contract RVS1Test -vv
Latch finality in the branch that moves genuinely-reserved bonus (the dust-donation griefing case goes through the other branch, which still does not latch):
Alternatively, on a good-faith CORRUPTED re-flag, reserve the already-swept bonus rather than re-snapshotting the drained totalBonus.
Impact – Medium Up to the entire bonus pool can be routed to the wrong party for good, with no on-chain way to unwind it, and in a stakerless pool the effect is total since bountyEntitlement computes to zero and claimAttackerBounty reverts outright. This impact is definitely not High: the pool stays solvent throughout with no principal ever at risk, and the money ends up at the sponsor's own recoveryAddress, which in most pools means a sponsor recovering a bonus they funded themselves. The whitehat's claim on that bonus also only exists because the moderator changed their mind after the fact. Likelihood – Low Four separate things have to coincide: nobody touches the pool for the whole active-risk window (or there are no stakers to begin with), the moderator flags SURVIVED and later reverses to CORRUPTED, that reversal is good-faith with a named attacker, and a sweep lands between the two flags. The first cuts against staker self-interest, since skipping the poke costs them their entire bonus share, and the second asks the moderator to overturn a scope judgement, which is a bigger deal than the typo fix DESIGN.md #4 offers the window for. The sweep itself I'd treat as near-certain once the rest holds, given it's permissionless and the sponsor has an obvious reason to make the call, but assembling the first three in one pool lifecycle is where this stays rare.
The contest is live. Earn rewards by submitting a finding.
This is your time to appeal against judgements on your submissions.
View preliminary resultsAppeals are being carefully reviewed by our judges.
The contest is complete and the rewards are being distributed.