External keeper/relayer not redundant
Midas's assessment for RD-F-062 — scored yellow on the v1.7.0 rubric. The evidence below is the curator's reasoning for this score.
Evidence summary #
Midas standard redemption process involves off-chain operational agents for T-bill sales and USDC return. The oracle NAV price update (setRoundData) relies on a single admin role (M_TBILL_CUSTOM_AGGREGATOR_FEED_ADMIN_ROLE) with no confirmed redundancy or fallback keeper. If the oracle admin fails to update prices, the CustomAggregatorFeed returns stale data indefinitely (no staleness reversion). Instant redemption vaults are on-chain but rely on oracle freshness.
Sources #
- GitHubSherlock 2024-08 README — Admin Centralizationsherlock-audit/2024-08-midas-minter-redeemer README: admin centralization acknowledged as known risk; instant vs standard redemption flowsretrieved 2026-05-16
Methodology #
Determine whether the protocol depends on a single keeper or relayer (Gelato, Chainlink Automation, custom) with no redundancy or failover.
See the full factor methodology and distribution across all protocols →