On June 9, 2026, risk firm LlamaRisk published a binding risk management framework for Aave — the largest decentralized lending protocol with approximately $14.5 billion in total value locked. The proposal, designated ARFC (Aave Request for Comments), introduces a four-layer risk taxonomy coverin...
"Over the past several weeks, Aave has been developing a new risk framework that includes asset risk, bridging risk, chain risk, and advanced automation capabilities for risk management. This framework establishes a new standard for how Aave assesses, monitors, and manages risk across the protocol." — Stani Kulechov, Founder, Aave
On June 9, 2026, risk firm LlamaRisk published a binding risk management framework for Aave — the largest decentralized lending protocol with approximately $14.5 billion in total value locked. The proposal, designated ARFC (Aave Request for Comments), introduces a four-layer risk taxonomy covering asset evaluation, bridge security, automated monitoring, and chain-level deployment criteria. If ratified by governance, it would apply across Aave V3, V4, and the forthcoming Aave Horizon institutional product.
The framework is a direct structural response to the $292 million KelpDAO bridge exploit of April 18, 2026 — the single largest DeFi hack of the year. That incident, attributed to North Korea's Lazarus Group, exposed a systemic vulnerability: an attacker minted 116,500 unbacked rsETH tokens through a compromised LayerZero bridge, deposited them as collateral on Aave V3, and borrowed $193 million in WETH against them. The resulting bad debt — estimated between $124 million and $230 million depending on loss socialization mechanics — triggered an $8.45 billion drawdown in Aave's TVL within 48 hours.
This report examines the framework's specific requirements, compares them against existing DeFi risk standards and traditional finance precedents, and assesses whether they are sufficient to prevent a recurrence. The data suggests a meaningful structural improvement, but gaps remain — particularly around cross-protocol contagion mapping and governance enforceability.
The sequence of events on April 18 exposed a chain of failures that the new framework directly targets.
KelpDAO operated a cross-chain bridge built on LayerZero's messaging infrastructure. The attacker exploited a vulnerability in the bridge's verification system — specifically, a 1-of-1 signer configuration that required only a single validation to authorize cross-chain token minting. Using this flaw, the attacker minted 116,500 rsETH tokens with no corresponding backing on the source chain.
The unbacked tokens were then deposited as collateral into Aave V3 markets on Ethereum and Arbitrum. Aave's price oracles treated the rsETH as legitimate collateral, enabling the attacker to borrow approximately $193 million in WETH. By the time the exploit was detected, the borrowed WETH had been extracted and the attacker's rsETH collateral was worthless.
At least nine DeFi protocols were directly affected. Aave's TVL fell from $26.4 billion to $17.95 billion in two days — a $8.45 billion decline, or 32%. Total DeFi TVL dropped $13.2 billion in the same period, according to CryptoBriefing.
The exploit highlighted three specific failure modes that the new framework addresses: insufficient bridge verification (1-of-1 signer), absence of automated exposure limits on newly bridged assets, and lack of standardized risk assessment for cross-chain collateral.
The KelpDAO incident did not occur in isolation. Through May 2026, cumulative DeFi exploit losses reached $840 million across more than 50 incidents — a 70% increase year-over-year from 30 incidents over the same period in 2025.
April 2026 was the single worst month in DeFi's recorded history by number of incidents, with more than 30 separate attacks netting approximately $635 million. The two largest:
Cross-chain bridges have become the dominant attack vector. Bridge exploits account for the majority of total dollar losses in 2026, continuing a pattern established by the Ronin ($625M, 2022) and Wormhole ($320M, 2022) incidents. The structural problem is unchanged: bridges aggregate large pools of locked assets behind relatively thin security perimeters.
LlamaRisk's framework organizes risk management into four distinct layers, each with specific quantitative requirements. Version 1.1 was published on June 12, 2026, with clarifications on bridge-stack role definitions following LayerZero feedback.
Layer 1 — Asset Risk. Governs the entire asset lifecycle: onboarding, quarterly due diligence refreshes, material-change re-evaluations, parameter adjustments, and deprecation. Key requirements include:
Layer 2 — Bridging Risk. Directly addresses the failure mode behind the KelpDAO incident:
Layer 3 — Monitoring and Automated Risk Oracles. Requires continuous automated surveillance with two specific mechanisms (detailed in the Automated Defense section below).
Layer 4 — Chain Risk. Evaluates whether Aave should deploy on a given blockchain at all, assessing network architecture, decentralization metrics, finality mechanics, governance control structures, and ecosystem adoption.
The framework introduces what amounts to a binding veto power for risk service providers — a structure without clear precedent in DeFi governance. Service providers hold explicit authority to block asset onboarding or force delisting under any of the following conditions:
Robby Greenfield, CPO of Bluprynt, described this hard-block taxonomy in the Aave governance forum as "the most important operational innovation" in the proposal, noting that it creates enforceable veto authority within a DAO governance structure — a function that typically requires centralized oversight in traditional finance.
Two automated systems operate on the Chainlink Runtime Environment and are owned by Aave DAO:
Automated Freeze Guardian: Halts reserves on adverse signals. When oracle data, on-chain activity patterns, or external alerts cross predefined thresholds, the system can freeze affected reserves without waiting for governance votes.
Supply and Borrow Cap Oracle: Automatically tightens protocol exposure by adjusting supply and borrow caps based on real-time risk signals.
Risk Steward parameters specify mandatory delays on parameter changes:
These delays are designed to prevent rapid parameter manipulation while still allowing automated response to acute threats through the freeze guardian.
The KelpDAO exploit's aftermath produced an unusual industry coordination effort. A coalition called "DeFi United," led by Aave, raised over $300 million in ETH commitments to restore rsETH's backing and recover approximately 107,000 rsETH in excess collateral.
Major contributors included:
| Contributor | Commitment | |---|---| | Mantle Treasury | Up to 30,000 ETH (loan) | | Stani Kulechov (personal) | 5,000 ETH | | EtherFi Foundation | 5,000 ETH | | Lido Finance | Up to 2,500 stETH | | Golem | 1,000 ETH |
The recovery plan involves a controlled liquidation sequence: temporarily adjusting the rsETH oracle price to enable efficient liquidation of eight affected positions, moving liquidated rsETH collateral to a DeFi United-controlled multisig, redeeming it for ETH through KelpDAO, and using the proceeds to repay deficits in Aave markets.
As of mid-June 2026, deployment of the recovery plan remains contingent on finalized agreements and governance approvals. LayerZero and KelpDAO have added new safeguards, but — as the recovery plan document itself states — "residual risk remains until those measures are validated in production."
The framework addresses the specific failure mode of the KelpDAO exploit but has identifiable gaps.
Cross-protocol contagion mapping. Kazuki Kaneshiro of ZKSC noted in the Aave governance forum that the framework does not address cross-protocol dependency mapping. During the KelpDAO incident, protocols with no direct rsETH exposure suffered losses because they shared WETH reserve pools with affected markets. "The shared WETH Reserve connected exposure with nothing to do with rsETH," Kaneshiro observed. The framework treats assets individually but does not model systemic interconnections across reserve pools.
Governance enforceability. A broader challenge identified by 21Shares research is that "token-holder ownership in DeFi is often based on norms and expectations, not legally or contractually enforceable rights." The risk framework creates operational standards, but its enforceability ultimately depends on governance participants' continued compliance. There is no external regulator or court that can compel adherence.
Aave Labs governance tensions. The risk framework arrives during a period of broader governance friction. Marc Zeller of the Aave Chan Initiative has questioned the bundling of governance decisions and demanded clearer revenue definitions. The February 2026 "Aave Will Win Framework" proposed routing 100% of product revenue to the DAO treasury, but the definition of "revenue" allows deductions controlled by Aave Labs with delayed disclosure — a structural tension between operational efficiency and governance oversight.
Retroactive scope. Existing listed assets that do not meet the new standards face removal, but the framework does not specify transition periods or grandfathering provisions for assets already accepted under older criteria. This creates potential market disruption risk if governance votes to delist assets retroactively.
The framework represents a convergence toward traditional finance (TradFi) risk management principles, adapted for on-chain implementation.
| Dimension | Aave Risk Framework | TradFi Equivalent | |---|---|---| | Asset evaluation | Quarterly due diligence, hard-block conditions | Bank credit committee, ongoing monitoring | | Bug bounty floor | $50,000 minimum | Vendor security assessments | | Bridge verification | 3-of-N minimum signers | Correspondent banking KYC | | Automated monitoring | Freeze guardians, cap oracles | Circuit breakers, position limits | | Veto authority | Service provider hard-blocks | Chief Risk Officer override | | Enforcement mechanism | Governance vote | Regulatory mandate |
The Ethereum Enterprise Alliance (EEA) has separately published DeFi Risk Assessment Guidelines (Version 1), which categorize risk across smart contract, economic, and liquidity dimensions. Contributors include OpenZeppelin, Consensys, EY, CertiK, and Quantstamp. Aave's framework is narrower in scope — focused on its own protocol — but more operationally specific, with quantitative thresholds rather than general principles.
The critical gap between DeFi risk management and TradFi remains enforcement. In traditional finance, risk standards carry regulatory backing — a bank's Chief Risk Officer operates under a legal mandate. In DeFi, risk service providers operate under governance delegation, which can be revoked by token vote.
Aave's four-layer risk framework marks the most operationally detailed attempt to systematize risk management in decentralized lending. The specific requirements — three-verifier bridge minimums, $50,000 bug bounty floors, 72-hour parameter delays — convert previously informal expectations into quantitative governance standards.
The framework directly addresses the failure mode that enabled the $292 million KelpDAO exploit. Had the three-verifier bridge minimum been in place, the 1-of-1 signer configuration exploited by the attacker would have been a hard-block condition preventing rsETH from being listed as collateral.
Whether the framework is sufficient depends on two factors that remain unresolved. First, cross-protocol contagion — the mechanism that amplified KelpDAO's damage across the DeFi ecosystem — is not modeled in the current framework. Second, governance enforceability remains dependent on token-holder compliance rather than legal mandate, leaving the framework structurally weaker than equivalent TradFi standards.
The proposal awaits governance ratification. Non-compliant assets would face removal in subsequent weeks. The outcome will determine whether DeFi's largest lending market operates under formalized risk standards or continues to rely on ad hoc assessment — a question with direct implications for the approximately $14.5 billion in user deposits currently held across Aave markets.