Audits & security
Audits
External security audits are in progress. Completed reports, including the auditing firm, scope, reviewed commit, and a summary of findings and their resolution, will be published here as they land.
Known issues and accepted risks
This section lists acknowledged properties of the protocol that auditors or community reviewers have flagged. Each entry includes the reason the property exists and the mitigation, if any.
Currently acknowledged:
- Multi-branch redemption reverts on any single-branch failure. Documented in Risks › Redemption revert. Mitigated by branch shutdown removing the failing branch from the basket.
- No fee-on-transfer collateral support. The protocol assumes ERC-20
transfermoves exactly the requested amount. Adding fee-on-transfer collaterals as branches is excluded by deployment policy. - Below-MCR troves can be carried for some time if no one liquidates them. Liquidation is permissionless and paid (200 RD gas compensation plus 0.5% of the trove's collateral), but when gas is expensive a small trove may wait, and the branch carries any shortfall until it is liquidated. Documented in Risks › Below-MCR trove not liquidated.
Bug bounty program
🐛 The bug bounty program launches with mainnet; the reporting channel and full rules will be published here.
Expected scope (subject to the program's published rules):
- In scope: the RAI Dollar contracts deployed on Ethereum mainnet at the addresses listed on Deployed addresses, including singletons (
Aggregator,RDToken,MOAIToken,RateParControl,MarketOracleAggregator,GlobalFeeRouter, etc.), the RD market pool hooks (RdStableSwapHook) and per-branch contracts. - Out of scope: front-end UIs, off-chain infrastructure, theoretical attacks on Chainlink, Chronicle or Uniswap v4 themselves, governance disclosures not yet announced, oracle staleness within published thresholds.
Severity guidance (typical for protocols of this kind, exact bounty schedule is set by the program):
| Severity | Definition |
|---|---|
| Critical | Loss of user funds, RD supply inflation outside the protocol's rules, or permanent freeze of major funds. |
| High | Loss of protocol revenue, partial freeze, recovery-mode bypass, or controller manipulation beyond stated bounds. |
| Medium | Incorrect accounting that can be unwound; griefing without permanent loss. |
| Low | Theoretical issues that require unrealistic conditions, dust-level accounting drift, missing event emissions. |
Responsible disclosure
If you've found a vulnerability:
- Do not disclose publicly.
- Submit through the private bug bounty channel once it is published, with a clear proof of concept. Until then, request a private reporting path through an official project account without including exploit details in the initial message.
- Allow reasonable time for a fix before public disclosure.
- The protocol team will coordinate disclosure timing after appropriate mitigations are in place.
Verification work
Separately from external audits, the protocol's development process includes:
- Formal verification with the Certora Prover: specifications covering the core contracts (borrowing, liquidation, redemption, the Aggregator, fee routing, staking, the price feeds and the RD market oracle).
- Stateful fuzzing with Echidna against multi-branch deployments.
- Static analysis with Slither, with the findings on the in-scope contracts triaged.
- Mutation testing of the contracts, to check that the test suite catches injected faults.
Results of this internal work are not published here yet.
Where to read code
- The protocol's source repository is currently private. Contract source is to be verified on Etherscan at deployment (see Deployed addresses).
- These docs reflect the mainnet deployment configuration.
- Audit-targeted commits will be tagged when audits begin.
Where to ask security questions
- A public security-discussion channel and a private vulnerability-reporting channel will be linked here when they are available.
- Do not post vulnerability details in public support or community channels.
What the protocol does not offer
- No insurance. There is no protocol-level insurance fund or backstop for SP depositor losses, oracle failures, or smart contract bugs.
- No upgrade path. Core contracts are immutable. If a critical bug is found, the response is branch shutdown + redirect of activity to a redeployed instance, not a contract upgrade.
- No admin recovery. No admin key can recover stuck funds, override liquidations, or change parameters on a live branch.
The trade-off is intentional: the protocol is auditable as a fixed artifact, with no surprise upgrades, but the cost is that severe issues require a fresh deploy and user migration.