The memo
REJECTED PROTOCOL-WIDE, WITH A MARKET-SPECIFIC REOPEN PATH; not a postponed judgment. The engineering case is real but narrower than the prior memo implied. Dolomite documents six audit firms across its inherited core and actively developed modules and reports 100% line, statement, and branch coverage; the same page warns that module audit scopes differ as integrations change. The core accounting layer is generally immutable, while mutable modules wrap external assets and strategies. That distinction is the file. A USDC lender is not simply taking stablecoin-credit risk: all deposits enter a virtual balance usable across lending, internal trading, margin, and integrated isolation-mode wrappers. Governance and operators can list assets, change risk parameters, approve global operators, and use specified emergency functions without the ordinary timelock. The current protocol has no located bad-debt event, but the project record is not exploit-free: a retired 2019 Loopring-era contract with stale user approvals was exploited for about $1.8M on 2024-03-20; 90% was recovered and the treasury reimbursed the rest. At the 2026-08-14 check, DefiLlama reports about $187.4M of TVL across chains and $304.8M of borrowed value, an adapter/accounting relationship this review has not reconciled to Dolomite's virtual-liquidity ledger. The audit count cannot substitute for underwriting each collateral, oracle, wrapper, caps, available cash, and admin path. No client allocation should proceed until a named USDC or USDT market dossier proves that its borrowers and liquidation routes exclude the exotic integration surface.
What would reopen the file
- Any bad debt exceeding the safety buffer
- Virtual liquidity permitting a DEX failure to reach lending depositors
- New collateral listings that cannot be partially liquidated
Facts on file
- Verdict
- Rejected
- Exposure
- stable lending
- Chains examined
- Arbitrum, Ethereum
- Instruments
- USDC, USDT, WETH
- Memo version
- v1
- Reviewed
- Next review