Summary: Dolomite is best understood as an internalized-liquidity margin and lending protocol rather than as a plain money market. Its key architectural move is the split between an immutable core margin engine and a mutable module layer, combined with subaccounts, internal liquidity transfers/trades, isolated borrow positions, and wrapper logic that can preserve some asset-native rights while collateral stays inside the system. The reusable mechanism insight is that Dolomite tries to merge trading, borrowing, and asset-specific integrations into one account-and-liquidity environment, while leaving real control concentrated in market listings, oracle policy, operator approvals, isolation wrappers, and module-level admin permissions.
What it does:
Supports overcollateralized borrowing, margin trading, spot trading, and related portfolio actions through the Dolomite Margin contracts
Uses an immutable core plus a mutable module layer so new features and asset integrations can be added without replacing the base margin engine
Lets users manage multiple isolated borrow positions through subaccounts, so one position can change without directly contaminating another
Uses internal or virtual liquidity, including internal transfers and trade actions, so users can borrow and trade inside the system without necessarily withdrawing ERC-20 liquidity from the protocol
Supports Isolation Mode wrappers and other rights-retaining integrations that let users keep exposure to certain asset-native behaviors, rewards, or vault flows while still borrowing against positions
Applies market-specific oracle choices, interest setters, supply/borrow caps, and margin/spread premiums on top of global risk settings
Key claims:
The docs say Dolomite combines the strengths of a DEX and a lending protocol and uses a two-layer architecture with an immutable core and a mutable module layer, which is the cleanest statement of what makes it analytically distinct from ordinary money markets
The main docs emphasize internal liquidity and subaccount transfers/trades, including the idea that a user can borrow into one subaccount and trade internally without withdrawing tokens from the system, which helps explain why Dolomite should be compared with internalized-liquidity systems rather than with simple lending pools
The developer overview confirms that Dolomite Margin is the core contract system behind the app and positions the protocol as modular while preserving immutable base behavior
The Isolation Mode docs say a borrow position can contain only one isolation asset, different isolation levels constrain what debt/collateral combinations are allowed, and isolation mode cannot be disabled once enabled, making wrapper and listing design real control surfaces
The admin docs show that core RiskLimits are immutable, but admins can still add markets, set oracles and interest setters, adjust global and market-level risk parameters, approve global operators, and in some cases bypass the timelock, so “immutable core” does not mean governance-light in practice
The listing docs say only the protocol administrator can list new markets today, while DAO voting rights and a DAO-appointed listing committee may later receive that power, which makes listing discretion central to Dolomite’s real governance surface
The repository README shows Dolomite inherits from and extends a dYdX-derived margin system, adding mechanisms for many markets, recyclable markets, operator controls, and other gas/risk-management changes; that lineage helps explain why Dolomite feels closer to a configurable margin engine than to a conventional pooled lender
Whitepaper: No single canonical whitepaper stood out in this pass. The strongest primary materials were the official site, docs root, risk/admin/listing docs, and the public protocol repository; see ../whitepapers/dolomite-primary-sources-2026-05-09.md.