The cover
ranking

The covered events (exploit, oracle, liquidation, governance) map closely onto lending risk, and Aave V3 is among the most audited protocols in DeFi, so the fit is good. The material limitations are in the exclusions: a GHO depeg is excluded because Aave issues GHO itself, losses from admin pauses are excluded, and cover bought after a public warning does not pay for that event.

The one covered event that truly maps to an AMM, a smart-contract exploit, sits against an unusually strong target: an immutable, non-upgradeable core audited by Trail of Bits and ABDK, with a $500k bounty and no core exploit since May 2021. So the covered peril is real but low-probability. The material limitation is fit: two standard covered events (oracle failure, liquidation) are lending concepts with no realistic trigger on an AMM, and the dominant economic risk of providing liquidity, impermanent (divergence) loss, is a market price movement and explicitly excluded. This cover protects the contract, not the position.

The covered events map onto Pendle’s own contracts well, and those contracts are strong: audited by Ackee, ChainSecurity, Spearbit, Dedaub and Code4rena, with no core exploit since 2021. The material limitation is structural. Pendle’s value derives from underlying yield sources (Ethena sUSDe, Aave, RWAs) and it hosts an ecosystem of protocols built on top; losses from those sit outside the wording. The 2024 Penpie hack ($27M) exploited a protocol built on Pendle, not Pendle itself, and illustrates the boundary. PT price moves before maturity are market movements, also excluded.

A notable mismatch between what is covered and where losses have occurred. Safe’s contracts are among the most battle-tested in crypto, formally verified, with no contract-level compromise since 2018, which is the risk this cover addresses. The two large losses routed through Safe, Bybit and WazirX, came from a compromised frontend and deceived signers, which §9.8 and §9.2 exclude. Buyers should be clear that the cover addresses the contract layer, not the operational layer where these losses originated.

Unlike most listings, Curve has proven both sides in practice: a code-level exploit really happened (July 2023, via a Vyper compiler bug that drained the CRV/ETH and other pools), and Nexus opened a formal claims process for cover holders. That makes the covered peril concrete. Two caveats stand out. First, the 2023 losses were largely recovered by whitehats and reimbursed by the teams, so it never became a large clean payout test, and it ran under the older Protocol Cover wording v1.0, not today terms. Second, crvUSD, Curve own stablecoin, is excluded from this cover under clause 9.6 and sold separately as a depeg cover, and a depeg of the stablecoins sitting in Curve pools is excluded as a market movement, which is a real gap for a stableswap venue.

A different animal from the retail listings: private, fixed-price at 0.75% per year, with a 5,000 ETH deductible, so it pays only on a catastrophic, correlated slashing event, not routine slashing. That structure fits its target, because restaking multiplies the slashing surface: on top of Ethereum consensus slashing, EigenLayer AVS slashing is now live and can even redistribute slashed funds. Ether.fi is a top-tier, heavily audited LRT (weETH carries an A+ risk rating), which supports the covered protocol. The caveats: the attachment point is very high (nothing until 5,000 ETH, tens of millions of dollars), it is not openly purchasable, the restaking-slashing surface it insures is genuinely new and largely untested, and non-slashing LRT risks (a contract bug, an eETH discount) are not this cover. It is also the single largest exposure in the whole set.

The covered smart-contract-exploit event fits an AMM, and Aerodrome has no core exploit on record. But three things temper it. It is young (August 2023, Slipstream since March 2024), so the track record is short. It is a double fork, of Velodrome / Solidly and of Uniswap v3, so the known-vulnerability clause reaches two upstream codebases it does not control. And it lives entirely on Base, one OP-Stack L2: a sequencer or chain-level fault is outside a protocol cover. As on any AMM, impermanent loss is excluded and the oracle and liquidation covered events barely apply. One relief: AERO is an emissions token, not a peg, so there is no depeg gap like Curve crvUSD.

Velodrome has run since June 2022 with no contract exploit of user funds, has Spearbit audits and an Immunefi bounty, and is the more battle-tested member of the ve(3,3) pair with Aerodrome. Its 2022 insider incident (a team member drained a team wallet, user funds untouched and returned) is a trust flag but not a contract failure, and notably it is exactly the kind of loss a protocol cover does not pay. The limits are the family ones: a fork lineage that clause 9.4 reaches into (Solidly, plus Uniswap v3 via Slipstream), everything on a single OP-Stack L2 (Optimism), impermanent loss excluded and the oracle and liquidation covered events barely applicable. VELO is not a peg, so there is no depeg gap.

Convex is a layer on Curve, and this cover protects the layer, not what is underneath. The listing spells that out itself: its metadata excludes losses that occur outside the designated protocol, such as a separate protocol used in an underlying yield aggregation strategy. Since Convex positions ultimately sit in Curve pools, a Curve exploit (as in July 2023) points a Convex holder to Curve cover, not this one. What remains genuinely covered is a bug in Convex own contracts, on a target with no core fund loss to date but a complex admin surface and a June 2022 frontend hijack. The three explicit exclusions make the boundary honest and clear; they also make the covered surface thin.

This is the right shape of cover for a synthetic dollar: it targets the headline risk, a depeg, directly and with a clear mechanical trigger, unlike a protocol cover that would exclude it. Two things temper it. The trigger is strict: NUSD must trade at least 10% below its reference value for seven consecutive days, so a sharp depeg that recovers within days, the common pattern, pays nothing. And the covered token is exotic: NUSD is backed partly by illiquid OTC-acquired locked tokens and by hedges on exchanges, so the peg leans on active management and off-chain custody. On the plus side, the audit roster is strong (Cantina, Spearbit, Sherlock) and the payout is a generous 0.975 per token, extended to sNUSD and the Pendle and Spectra PT wrappers.

The wording covers exploits, oracle failures and governance attacks on the ether.fi contracts, and the protocol is well audited with no major incident to date. Three limitations matter for a restaking product: validator slashing is not a covered event here (Nexus sells a separate slashing product for that), an eETH/weETH depeg is excluded, and losses cascading from the EigenLayer layer underneath are undefined in scope. Notably, no active cover currently exists on this listing.

This is the riskiest listing in the set so far, and the cover fits it least well. What is covered is an exploit of InfiniFi own contracts, on a protocol with genuine strengths (renounced ownership, fully on-chain reserves, tests and fuzzing, a coded loss waterfall) but real weaknesses (new in 2025, novel design, no named third-party audit we could confirm, and reserves deployed into higher-risk venues including Ethena). The two dominant risks sit outside the cover: an iUSD depeg is excluded as a protocol-issued asset (9.6), and losses at the underlying farms are outside the designated protocol. The market agrees it is riskier: cover here is priced near 2.6% per year, roughly nine times a blue-chip AMM.

The scope is unusually broad: five designated protocols, and the annex even re-includes the Bridge, which most covers exclude. On a perps venue the oracle and liquidation covered events genuinely fit, unlike on an AMM. But the annex carves out the two most likely loss modes: a Bridge multisig or MPC failure (the main way a bridge is lost) is excluded, and the HLP vault is covered only when it does not operate as intended, so the vault losing money on a bad position is out. The March 2025 JELLY incident showed exactly this: a manipulated market forced a ~$13.5M HLP loss, and a small validator set reached quorum in about two minutes to override the oracle and settle positions. Neither the vault loss nor the validator intervention is a covered exploit. At roughly 5% per year, this is the most expensive listing in the set: the underwriters own verdict on the risk.

A strong track record on a product that is no longer offered. Custody Cover has the best-documented paid-claims record in the mutual (FTX, BlockFi, Hodlnaut in 2022, with itemised decisions), but Nexus has retired the entire CEX custody class: no listing is buyable, capacity is zero. For exchange-custody protection today it offers none. The withdrawal is itself an informative signal: a provider that had paid on this risk chose to stop underwriting it, consistent with how correlated custodian failures proved to be in 2022.