The new Aave risk framework proposal from LlamaRisk
The framework systematically quantifies and monitors risk across four layers — assets, bridges, automated oracles, and chains — to set and continuously adjust Aave's lending parameters (LTV, caps, collateral eligibility) so that protocol exposure never exceeds what on-chain evidence supports.
LAYER 1: ASSET RISK
- Asset classification: governance-ratified allowlist mapping, comparable asset identification
- Multi-chain evaluation: bridge topology, smart contract divergence, oracle paths per chain
- Smart contract audits: firm reputation, deployed version coverage, unresolved critical/high findings, incident history
- Bug bounty coverage: program existence, $50k minimum payout floor, bounty-per-million-TVL scaling, scope coverage
- Liquidity & market structure: secondary market depth vs largest borrower, liquidation bonus/slippage, LP diversity, venue footprint, holder concentration, peg deviation (1% over 2-day window), volume profile
- Timelock requirements: timelocks on parameter changes, mint/burn authority, oracle authority, bridge authority, upgrade paths; delay sufficiency (Level 0-5)
- Signing authority decentralisation: signer-set composition, quorum thresholds, identity verification, custody standards; MPC-specific: shard count/distribution, custody framework, open-source crypto library
- Legal disclosures: issuer jurisdiction, redemption rights, asset-backing terms, custodian arrangements, regulatory status
- Holder claims seniority: holder class identification (canonical, bridged, wrapper, custodied), seniority ordering, cross-chain symmetry
- Asset backing structure: proof-of-reserves attestation, reserve composition, on-chain visibility, attestation cadence, backing-asset price observability
- Issuer operational stack: hosting, key custody, incident response, change management, third-party dependencies
- Incident communication: escalation paths, 24/7 reachability, pre-notification for material changes
- Continuous due diligence: quarterly refresh, delta reports, per-chain coverage re-evaluation
- Material change taxonomy: tracks ~30 change types including new mint/burn authorities, MPC/multisig changes, collateral type changes, oracle adapter swaps, feed swaps, rate-limit changes
- Deprecation triggers: liquidity decay, oracle stability degradation, feed manipulability, thin feed conditions, Chainlink risk tier flags, backing impairment
- Oracle treatment: Chainlink risk tier, internal oracle path identification, flash-loan deviation testing, exchange-rate monotonicity, single-transaction manipulation resistance
LAYER 2: BRIDGING RISK
- Bridge topology: origin chain, canonical supply, representation type (mint-and-burn / lock-and-mint / native), verifier configuration per route
- Verifier set properties: independent verifier count, operator diversity (organizational, geographic, infrastructure), supply-chain isolation
- Library & configuration: receive library pinning, upgrade path control, route topology
- Bridge authority timelocks: timelocks on ownership transfer, verifier-set changes, library upgrades, rate-limit changes, mint/burn authority grants
- Pause pathway: issuer-side pause, vendor halt capability, Aave-controllable pause, multi-path documentation
- Bridge-layer custody: institutional-grade custody verification, key custody assessment, change pre-notification
- Rate limiting: native or external rate-limiting per route, inbound/outbound limits, sizing based on highest sustained flow headroom
- Incident response: 24/7 redphone across vendors/attestors/issuers, written exploit procedures, rehearsal records
- Monitoring teams: dedicated security teams for bridge stack and issuer, automated anomaly detection, pause authority
- Bridge configuration lifecycle: per-lane/per-route/per-direction initial limits, comparable issuer references, adjustment documentation
- Bridge stack evaluation: vendor operational security, code quality, audit reports, exploit history/recurrence, remediation quality
LAYER 3: MONITORING & AUTOMATED RISK ORACLES
- Continuous monitoring: authority-queue activity, governance activity, infrastructure changes, direct asset-state (bridge balance reconciliation, backing-asset state, minting/burning anomalies)
- Automated Freeze Guardian triggers: backing-asset price signals, reserve composition, redemption-buffer replenishment, exchange-rate manipulation, locked-vs-minted supply mismatches, unannounced verifier/library changes
- Supply/Borrow Cap Oracle triggers: secondary market thinning, holder exit velocity, cross-chain supply concentration, proportional cap adjustments
- Risk Steward parameters: minDelay settings (36h for caps/rates, 72h for collateral/E-Mode, 48h for Pendle discount), maxPercentChange bounds
- Umbrella coverage: unstaking cooldown sizing, freeze conditions under asset-level stress, upstream signal tracking, bad debt recognition timeline
LAYER 4: CHAIN RISK
- Network architecture: consensus mechanism, finality model, settlement layer, EVM divergence, audit coverage, bug bounty
- Decentralisation: validator/sequencer count and identity, geographic and client diversity, staking economics, stake concentration
- Finality & withdrawals: finality lag, reorg history/depth, L2 withdrawal delays, forced-inclusion/escape-hatch availability, fraud-proof/zk-proof maturity, MEV characteristics
- Chain governance: upgrade authority, multisig composition/quorum, timelock delays, pause/emergency authority, governance token distribution, voting concentration
- Operational history: halt/reorg incidents, post-mortems, time-to-recovery, recurrence patterns
- Ecosystem adoption: aggregate TVL, stablecoin TVL/dominance, active addresses, protocol count/concentration
- DeFi vertical coverage: lending, DEX, stablecoins, derivatives, oracle infrastructure gaps
- Liquidity infrastructure: DEX depth on major pairs, bridge support, on/off-ramp corridors, CEX withdrawal corridors, competing lending market parameters
- Tooling & monitoring: block explorer parity, indexing platform support, RPC provider diversity, risk-monitoring tooling availability
- Native/major asset liquidity: native token DEX volume and price stability, major asset (ETH, BTC, stablecoins) depth, organic vs incentive-driven volume
- Per-asset deployment constraints: chain-tier upper bounds on LTV, supply cap, borrow cap; cross-chain expansion constraints; quarterly tier refresh
CROSS-CUTTING
- Remediation timelines: 1-month implementation target, hard-block triggers for immediate review, backlog persistence tracking
- Risk classification: asset-to-chain tier mapping, exposure tier constraints, parameter stack alignment, E-Mode configuration, comparable asset benchmarking
LlamaRisk has published an ARFC proposing a new standardized Risk Framework that governs all assets on
@Aave V3, V4, and Aave Horizon.
The framework establishes standards for evaluating asset, bridge, and chain-level risk criteria, and for monitoring and automated risk management systems.