Spark Savings V2 Onchain

Identification — confirmed protocol: Spark Savings V2 (Spark Vaults V2), official website spark.finance; documentation at docs.spark.finance. Category: ERC-4626 yield-bearing savings vaults / DeFi asset management. Yield is generated through the Spark Liquidity Layer, with rates managed by Spark Governance.

Ethereum is the requested and confirmed chain; initial Ethereum launch authorization was September 22, 2025, with the launch execution recorded in the October 6, 2025 Sky Governance action. Native token: No dedicated Spark Savings V2 native token identified. The products issue vault-share tokens: spUSDC, spUSDT, and spETH. SPK is an ecosystem/governance token, not the vault’s native accounting asset.

The underlying assets are USDC, USDT, and WETH respectively. Ethereum contracts and verification:

• spUSDC: 0x28B3a8fb53B741A8Fd78c0fb9A6B2393d896a43d — listed in Spark integration documentation and Sky Governance; proxy source verified on Etherscan.
• spUSDT: 0xe2e7a17dFf93280dec073C995595155283e3C372 — cross-checked in Spark documentation, Sky Governance, and Etherscan; proxy source verified.
• spETH: 0xfE6eb3b609a7C8352A241f7F3A21CEA4e9209B8f — cross-checked in Sky Governance and Etherscan; proxy source verified.
• SavingsVaultIntents: 0x592B7DB9906E6f8924C4D74c2A0aB86CE44fDDDf — documented by Spark; explorer verification not independently confirmed here. Dune requirement: Not verifiable as of October 10, 2026. Dune MCP was unavailable; therefore no on-chain Dune query ID/execution ID or raw-chain confirmation is provided. Fork lineage: Yes. Spark Vaults V2 explicitly identifies itself as a fork of Sky’s sUSDS. Changes include OpenZeppelin role-based access control replacing rely/deny, new admin/setter/taker roles, removal of DSS integration, take() liquidity deployment, UUPS upgradeability, and additional accounting helpers. Auditing / malicious-fork history: ChainSecurity completed a Spark Vaults V2 audit on October 1, 2025; it reported high assurance with informational issues/notes. Whether every deployed proxy implementation exactly matches the audited commit is Not verifiable as of October 10, 2026. Malicious modifications in similar forks: Not verifiable as of October 10, 2026.

Spark Savings V2 Onchain looks like a real, functional product rather than a pure landing page: the docs describe ERC-4626 vaults, standard deposit/withdraw flows, and an app path at app.spark.finance/savings for depositing accepted assets and obtaining vault tokens. The docs also describe live withdrawal handling, including instant redemptions within a liquidity buffer and asynchronous fulfillment for larger withdrawals on Ethereum Mainnet, which is consistent with a mature product workflow. The presence of a machine-readable OpenAPI specification at api.spark.finance/openapi.json is strong evidence of an open API, and the docs explicitly label it the Savings Data API Reference.

The API description says it returns current APY, TVL, deposit cap, liquidity, collateral composition, and token metadata, which indicates an externally consumable data interface rather than a closed-only UI. Documentation quality also appears solid: there are dedicated tutorials, integrator guides, and protocol/dev pages covering deposits, withdrawals, intents, and contract interfaces. I did not find clear evidence of broken links or fake-template signs in the retrieved sources.

Not verifiable as of 2026-10-10. There is one naming/branding inconsistency to note: some sources point to spark.finance or app.spark.fi while the docs reference app.spark.finance/savings, which suggests rebranding or multi-domain routing rather than a clear red flag by itself. Overall, the product appears mature, live, and API-enabled, with withdrawal functionality that extends beyond a static marketing site.

Spark has an active bug bounty program managed through Immunefi. The published scope includes Spark Savings V2 contracts, with critical smart-contract findings rewarded at 10% of directly affected funds up to a maximum of $5,000,000; high-severity smart-contract findings are capped at $100,000. A proof of concept is required for critical and high smart-contract issues, and rewards are paid in DAI on Ethereum, denominated in USD.

The program page states it has been live since 01 November 2023 and was last updated on 18 September 2026. No public bug-fix or payout results were found in the retrieved sources, so results are not verifiable as of 2026-10-10.

Scope: Ethereum deployment of Spark Savings V2 Onchain (spark-oc). Dune/on-chain verification was unavailable; therefore exposure percentages and current balances are not measurable. Dependency map and risks

• Primary dependency — Sky/USDS: V2 USD vaults (spUSDC, spUSDT, spPYUSD) are described as fully backed by USDS; deposits share common collateral portfolios and may be converted into another stablecoin. A USDS depeg, impaired redemption, governance error, or Sky collateral loss can transmit directly to all three vaults.
• Yield/allocation layer: Spark Liquidity Layer (SLL) allocates across DeFi, RWA products, and yield-bearing stablecoins, with off-chain monitoring and governance-approved strategies. This creates correlated smart-contract, manager, liquidity, valuation, and withdrawal-gating risk. Credora confirms common credit exposure across the USD savings vaults.
• External protocols/assets: Spark publicly identifies exposure routes involving Aave, Morpho, Ethena, Pendle, Maple/SyrupUSDC, Superstate USTB, BlackRock BUIDL, and Centrifuge/JTRSY. This list is an unverified marketing claim for current allocation size; exact Ethereum exposure is Not verifiable as of October 10, 2026.
• RWA issuer/SPV/custody risk: Tokenized-T-bill and credit products add issuer, fund, transfer-agent, legal-claim, NAV, redemption, and custodian risks. Centrifuge uses hub/spoke accounting and cross-chain synchronization; failure or gating at an issuer/SPV can delay or impair redemptions.
• CEX/MM exposure: Any Ethena USDe/sUSDe allocation inherits delta-hedging, centralized-exchange liquidity, and off-exchange-settlement/custodian risk. Ethena confirms backing depends on custodial providers and centralized derivatives venues.
• Oracles/manipulation: SparkLend uses a three-source median (RedStone, Chainlink, Chronicle), reducing single-oracle failure risk; however, stale, manipulated, or divergent prices can still cause incorrect valuation/liquidation. RWA/NAV and yield-token pricing introduce additional non-spot valuation risk.
• Bridges: The Ethereum vault itself is native Ethereum, so a user bridge is not required. However, SLL’s cross-chain allocations and tokenized-asset infrastructure can introduce bridge/message-passing risk. Amount at risk is Not verifiable as of October 10, 2026.
• Failure scenarios: USDS/USDC/USDT/PYUSD depeg; Ethena hedge or custodian failure; RWA issuer insolvency or redemption suspension; DeFi exploit/oracle attack; liquidity run. Spark documentation states withdrawals may be halted in recovery mode, creating exit/lock-up risk. Structured fields
• dependency_failure_active: false (no active dependency failure identified in reviewed sources as of October 10, 2026)
• max_exposure_pct: null (Not verifiable as of October 10, 2026)

Spark Savings V2 is structured as a non-custodial vault system: users deposit into ERC-4626 vaults, and redemptions are designed so the contract itself does not take custody of user shares or redeemed assets in the withdrawal-flow contract. Instead, withdrawal requests are fulfilled atomically by calling the vault to redeem directly to the recipient. For the custody question, the clearest documented answer is that custody remains with the smart contracts and the user-controlled recipient flow, not with a discretionary operator; the Spark Liquidity Layer and its permissioned taker/relayer roles manage liquidity deployment and fulfillment, but the docs describe the system as non-custodial.

On segregation of assets, the available documentation says each USD-denominated deposit is backed 1:1 by USDS and that the products are organized as separate vaults, but it does not clearly verify legally segregated client asset accounts in the traditional custodial sense. The safest statement is that asset backing is vault-based and protocol-level, while legal segregation of assets is Not verifiable as of 2026-10-10. withdrawal_paused: null segregated_assets: null

Key management for Spark Savings V2 on Ethereum is organized as role-based smart-contract access control, not as a single externally owned admin key.

• Default administration: Each V2 vault uses OpenZeppelin AccessControl and UUPS upgradeability. DEFAULT_ADMIN_ROLE can upgrade the implementation, grant/revoke roles, and set the allowed Vault Savings Rate (vsr) bounds. The Ethereum launch governance records assigned this role to the Spark SubDAO Proxy (0x3300…af8c4), making Sky/Spark governance the ultimate control layer.
• Rate-setting authority: SETTER_ROLE can change the active vsr within the admin-defined bounds. At launch this was the Spark Operations Safe (0x2E1b…eDfC); a later governance proposal specified migrating the role for Ethereum spUSDC, spUSDT, and spETH to the ALM Proxy Freezable (0x9Ad8…5178), which adds operational freeze capability.
• Liquidity deployment: TAKER_ROLE is assigned to the Ethereum Spark ALM Proxy (0x1601…347E). It can call the vault’s liquidity-removal function (take) to deploy assets through the Spark Liquidity Layer and return liquidity when needed. This is an operational custody/asset-movement role, distinct from governance administration.
• Large-withdrawal intents: The separate Savings Vault Intents contract uses DEFAULT_ADMIN_ROLE for configuration and a RELAYER role for fulfillment. The relayer is designed to execute withdrawals, but cannot redirect funds arbitrarily; the contract redeems directly to the user-designated recipient. Risk view: The architecture separates upgrade/governance, rate-setting, and liquidity execution. However, the public sources reviewed do not establish the current signer count, threshold, hardware/MPC practices, or a definitive timelock for each controlling proxy/safe. Not verifiable as of October 10, 2026. On-chain role membership and execution status are likewise not independently verifiable in this run because Dune is unavailable.

Scope: Ethereum Spark Savings V2 vaults: spUSDC 0x28B3a8fb53B741A8Fd78c0fb9A6B2393d896a43d, spUSDT 0xe2e7a17dFf93280dec073C995595155283e3C372, spETH 0xfE6eb3b609a7C8352A241f7F3A21CEA4e9209B8f, spPYUSD 0x80128DbB9f07b93DDE62A6daeadb69ED14a7D354; SavingsVaultIntents: 0x592B7DB9906E6f8924C4D74c2A0aB86CE44fDDDf. Architecture: ERC-1967 UUPS proxy → SparkVault implementation. DEFAULT_ADMIN_ROLE authorizes upgrades, deposit-cap changes, VSR bounds, and role management; SETTER_ROLE changes VSR within admin-set bounds; TAKER_ROLE calls take() to remove idle assets for Spark Liquidity Layer strategies. Users use ERC-4626 deposit/withdraw/redeem. Large redemptions use SavingsVaultIntents → ALM Planner/RELAYER → vault redemption. ``text User ──deposit/redeem──> SparkVault proxy ──> implementation │ TAKER_ROLE │ take() v ALM / external strategies │ transfer assets back v SparkVault liquidity Large withdrawal: User → SavingsVaultIntents → RELAYER/ALM → SparkVault.redeem() ` Admin and exit risk: Documentation identifies the Ethereum default-admin holder as Spark Subdao Proxy 0x3300f198988e4C9C63F75dF86De36421f06af8c4; operations setter is 0x2E1b01adABB8D4981863394bEa23a1263CBaeDfC, and ALM/taker is 0x1601843c5E9bC251A3272907010AFa41Fa18347E`.

These role assignments are not independently on-chain verified in this run. There is no documented pause/freeze or fee/oracle function in SparkVault V2. Direct withdrawals remain possible only up to idle vault liquidity; otherwise users depend on ALM liquidity preparation and RELAYER fulfillment.

Admin can set depositCap = 0; a malicious upgrade could alter accounting, withdrawals, or asset routing. TAKER_ROLE can remove all currently idle assets, but cannot exceed the vault balance. Critical gap: Dune MCP was unavailable. Therefore proxy-admin type, decoded admin events, current role holders, renouncements, and an on-chain timelock delay are Not verifiable as of October 10, 2026. No governance-enforced delay should be assumed.

Compromised upgrade authority is the worst case: arbitrary implementation, theft or permanent freezing of funds. Users cannot rely on immediate exit when liquidity is deployed externally. ChainSecurity’s V2 audit reports 0 critical and 0 high findings; this is an audit of the codebase, not proof that every deployment matches the audited artifact. Fields: admin_can_drain=true (conditional on compromised upgrade authority); audited_deployment=null; upgradeable=true; unresolved_critical=0; unresolved_high=0.

ChainSecurity audited Spark Vaults V2 for SparkDAO on Ethereum. The report covers SparkVault V2, an ERC-4626 vault with manually controlled Vault Savings Rate (VSR), whitelisted takers, UUPS upgradeability, and liquidity deployment via the Spark ALM Controller. Scope: src/ISparkVault.sol and src/SparkVault.sol, latest reviewed commit 0a686ba2fcf874bc1542171a323779ba73ac2dc5; dependencies, underlying tokens, and usage were excluded.

The report identifies asset-solvency and circular-reinvestment risks as notes, plus ERC-4626 preview-function liquidity behavior. Bytecode-match note: deployed Ethereum bytecode was not independently matched to the audited commit; Not verifiable as of 2026-10-10.

Assessment — Spark Savings V2 Onchain (Ethereum): The product is attributable to the broader Spark Protocol, developed by Phoenix Labs within the Sky ecosystem (formerly MakerDAO), rather than to an entirely anonymous team. Sam MacPherson is publicly identified as Phoenix Labs co-founder and CEO; he previously worked in MakerDAO’s protocol engineering team and is described as having prior smart-contract and blockchain-development experience. Other publicly associated Phoenix Labs co-founders include Nadia Alvarez, formerly MakerDAO’s Head of Growth, and Lucas Manuel; however, Lucas’s detailed professional background is not consistently disclosed in the available sources.

• Prior projects and outcomes: Spark launched in 2023 as a MakerDAO-originated lending protocol and expanded into Spark Savings and related capital-allocation products. The available evidence supports genuine prior DeFi engineering and deployment, not merely a marketing website.
• Security history: Search results identify audits of SparkLend and Spark Vaults v2. One third-party compilation reports no critical findings in a 2025 Spark Vaults v2 review, while separately identifying historical oracle-related critical/medium findings in SparkLend-related reviews; these are audit findings, not evidence of an exploit. A confirmed prior hack or loss attributable specifically to Spark Savings V2 is Not verifiable as of October 10, 2026.
• Public vs anonymous: Core leadership is publicly named, but the full team, beneficial ownership, legal entities, and accountability structure are not fully transparent in the evidence reviewed. Credibility is therefore stronger at the contributor level than at the corporate-ownership level.
• Office and jurisdiction: A Panama registered-office reference appears in an aggregator’s protocol record, but it is not independently corroborated here. A real operating office, domicile, and onshore/offshore status are Not verifiable as of October 10, 2026.
• Reality check: The protocol has credible signs of a real engineering operation—named contributors, Maker/Sky lineage, deployed products, and external audits—but the evidence does not establish a fully transparent legal business or independently verified physical headquarters. Claims about institutional scale or managed capital should not be treated as verified here.

Reputation assessment: mixed-positive, with material governance and counterparty risks. Spark Savings V2 is associated with Spark/Sky, the ecosystem formerly linked to MakerDAO; the available evidence identifies Spark Vaults V2 as an ERC-4626 savings-vault system. This identity is consistent with the Ethereum Spark protocol, but the requested contract-level on-chain verification is unavailable: Not verifiable as of October 10, 2026.

• Audits/security: Publicly indexed materials identify ChainSecurity and Cantina reviews covering Spark Vaults/savings, PSM, oracle, and related components. One aggregator reports a prior critical finding in a weETH oracle review and a medium rsETH pricing issue; it also reports no critical findings in a later Vaults V2 review. Audit scope is component-specific and does not establish that every deployed module is audited.
• Track record: Available independent commentary describes Spark as operational for more than two years without a documented exploit, but this is not equivalent to proof of safety.
• Founders/investors: Spark’s upstream association with Sky/MakerDAO is documented, but reliable independent identification of a distinct Spark founding team or institutional investor roster was not established. Not verifiable as of October 10, 2026.
• Sentiment and criticisms: The principal criticism is governance centralization: commentary attributes substantial influence over Spark’s risk parameters and token policy to Sky governance and Rune Christensen. Other concerns include exposure to Ethena USDe/sUSDe strategies, where depeg risk and negative funding rates could impair returns or capital. These are risk assessments, not allegations of fraud.
• Fraud, rug, insolvency: No credible source located in this review substantiated fraud, rug-pull, or insolvency allegations against Spark Savings V2. Not verifiable as of October 10, 2026 for a definitive negative finding.
• Legal/regulatory/sanctions: Public evidence reviewed did not establish a specific enforcement action, sanctions designation, or licensing determination for Spark Savings V2. Not verifiable as of October 10, 2026. > Unresolved concerns: incomplete audit coverage; centralized upstream governance; strategy/counterparty exposure; limited publicly surfaced KYC/KYB and sanctions controls; and the inability to reconcile claimed TVL, reserves, or exposure with raw Ethereum data. On-chain metrics and claimed figures are therefore Not verifiable as of October 10, 2026.

Scope: Spark Savings V2, Ethereum only; on-chain verification was unavailable in this run. Therefore: Not verifiable as of October 10, 2026 for contract-level balances, historical flows, revenue attribution, leverage, and Dune-vs-DeFiLlama reconciliation. Strategy/assets: ERC-4626 V2 vaults accept USDC (spUSDC), USDT (spUSDT), PYUSD (spPYUSD), and ETH (spETH). Deposits receive transferable vault shares whose redemption value increases with yield.

V2 assets are allocated by the Spark Liquidity Layer (SLL) across DeFi lending, RWA/tokenized-fund strategies, and yield-bearing stablecoins; governance controls rates and deployment. Stablecoin vaults are generally market-neutral in deposit denomination but retain USDS collateral, counterparty, liquidity, smart-contract, and RWA exposure. spETH is directionally exposed to ETH price risk. No restaking was identified in the reviewed sources; looping/leverage is Not verifiable as of October 10, 2026. Yield sustainability/subsidies: Yield is described as strategy-generated rather than a stated token-emission APY; Spark Points may be an additional incentive, but their economic value is not quantified.

Organic-yield share and subsidy share are Not verifiable as of October 10, 2026. Current DeFiLlama average Spark Savings supply APY is 3.34%; spUSDC Ethereum is 3.80%, with “stable” 30-day stability. Historical APY volatility is Not verifiable as of October 10, 2026. Lockups/withdrawals/limits: No lockup.

Native deposits/withdrawals are designed without slippage and generally use an idle liquidity buffer; larger Ethereum spUSDC/spUSDT withdrawals may use asynchronous Liquidity Intents, usually fulfilled within minutes. The documented buffer is up to $10m; vault-specific caps and current limits are Not verifiable as of October 10, 2026. Protocol fee beyond Ethereum gas is stated as none under current parameters. TVL: DeFiLlama reports Spark Savings at $1.389b total, $1.056b Ethereum (76%), +12.7% over 30 days; spUSDC Ethereum is $313.78m.

Other V2 product TVL and trend decomposition are Not verifiable as of October 10, 2026. Spark Savings-specific protocol revenue is Not verifiable as of October 10, 2026. Contradiction callout: DeFiLlama aggregates all Spark Savings products, while the requested slug is V2 Ethereum-only; treating $1.389b as V2 Ethereum TVL would overstate exposure.

Scope: Ethereum Spark Savings V2 comprises at least spUSDC, spUSDT, and spETH. Canonical Ethereum vault addresses identified from Spark documentation are spUSDC 0x28B3a8fb53B741A8Fd78c0fb9A6B2393d896a43d and spUSDT 0xe2e7a17dFf93280dec073C995595155283e3C372; spETH is 0xfE6eb3b609a7C8352A241f7F3A21CEA4e9209B8f. Size / composition: Spark states that spUSDC, spUSDT, and spPYUSD are backed by USDS and that yield is generated through the Spark Liquidity Layer (SLL). spETH is described as backed by Spark’s balance sheet and insurance. The underlying deposit assets are USDC, USDT, and ETH respectively.

These are protocol claims, not independently verified reserve balances. Reported exposure: Spark’s data dashboard recently displayed approximately $284.66M for spUSDC, $362.12M for spUSDT, and $71.80M for spETH. This is an analytics-platform view of vault balances, not a reserve attestation and not an on-chain-verified result. Custody / control: Assets are not necessarily retained inside the vault contracts. The SLL allocates capital across external strategies.

ChainSecurity describes the ALMProxy as holding funds and interacting with external contracts, with Controllers and rate limits governing operations; administrative roles can configure permissions. This creates governance, controller, allocator, counterparty, and liquidity risks. Reserve policy / attestations: Documentation describes liquidity management and risk controls, but I found no independent reserve attestation establishing total assets, asset-by-asset custody, or liabilities for Ethereum V2. The available ChainSecurity material is a code assessment, not a proof-of-reserves report.

Dune was unavailable in this run; therefore balances, treasury addresses, control relationships, composition, and liabilities are Not verifiable as of October 10, 2026. Reported figures above should not be treated as audited reserves.

Scope: Spark Savings V2 is a set of ERC-4626 receipt tokens (spUSDC, spUSDT, spETH, etc.), not a separately tokenized DAO. Its ecosystem token is Spark (SPK); the vault share tokens are not governance tokens.

• Token/address: SPK, Ethereum ERC-20: 0xc20059e0317DE91738d13af027DfC4a50781b066.
• Supply/valuation: Genesis supply and stated maximum are 10B SPK. CoinMarketCap reports 3.084B circulating, market cap $66.9M, and FDV approximately $216.8M as of October 10, 2026; Spark Data Hub reports materially different figures—3.352B circulating, $76.5M market cap, $228.2M FDV. Contradiction: circulating supply and valuation are not consistent across sources; no Dune/on-chain reconciliation is available.
• Utility/governance: SPK supports Snapshot signaling/sentiment checks, staking, and SPK farming. Staking currently advertises Spark/Symbiotic points rather than a documented protocol-revenue yield. Savings-vault rates are managed through Spark/Sky governance, not sp-token voting.
• Revenue share/buybacks/burns: No documented revenue-sharing mechanism or automatic buyback/burn program was found. The MiCA paper states no algorithmic minting, rebasing, or burning; Sky reportedly retains emergency mint authority. Not verifiable as of October 10, 2026 whether any announced unlocks, buybacks, or burns actually settled on-chain.
• Emissions/unlocks: Sky farming receives 6.5B SPK over 10 years: 1.625B/year in years 1–2, then declining to 203.13M/year in years 7–10. Ecosystem: 23% / 2.3B; team: 12% / 1.2B, with a 12-month cliff followed by three-year vesting.
• Concentration/insiders: Spark Data Hub reports top balances of 47.88% (Sky Pause Proxy), 9.54% (Spark subDAO proxy), and 7.29% (unidentified address). Team/treasury addresses are disclosed, but beneficial ownership and live balances are Not verifiable as of October 10, 2026 without raw on-chain reconciliation.
• Admin controls: Verified ABI exposes auth-gated mint, burn, rely, deny, and file; no blacklist or transfer-fee switch is shown. Active wards controllers are Not verifiable as of October 10, 2026.
• Liquidity/listings: Uniswap V3 SPK/WETH is the clearest Ethereum DEX listing. DEX Screener reports $131.2M liquidity, but also shows only <$1 WETH paired against 4.99B SPK—an internal inconsistency, so usable depth is Not verifiable as of October 10, 2026.

For Spark Savings V2 on Ethereum, a BTC move below $10,000 is best treated as an *extreme tail-risk macro shock* rather than a direct protocol-specific trigger. The documented user-facing risk is mainly depeg risk for USDS/sUSDS and smart contract risk, while Spark’s own risk framework says losses are intended to be absorbed first by the Sky Surplus Buffer, then Sky protocol capital reserves, then the SKY backstop, before reaching broader depositor exposure. Under a severe BTC crash, the most relevant transmission channels are indirect: broader crypto market stress, stablecoin depeg pressure, and potential liquidity impairment.

Spark states that in severe stress it can place Savings vaults into recovery mode, temporarily adjusting withdrawals to support orderly liquidity management and avoid bank-run dynamics. Spark Vaults V2 also document that withdrawals are constrained by available vault liquidity, and insufficient-liquidity reverts if the vault cannot satisfy the request. What is not verifiable as of 2026-10-10 from the provided sources is the exact loss path, insolvency probability, or the Ethereum-specific TVL/exposure that would be affected by such a BTC drawdown.

The available sources do not provide on-chain balances or a chain-specific exposure breakdown, so a quantitative stress loss estimate cannot be supported here. The most defensible scenario statement is: if BTC fell below $10,000, Spark Savings V2 on Ethereum would likely face heightened redemption pressure and possible temporary withdrawal restrictions in a broader market stress event, but any actual depositor loss would depend on the severity of downstream USDS/liquidity impairment and the effectiveness of the loss-absorption waterfall.

For Spark Savings V2 on Ethereum, a 20% depeg of the largest collateral is not verifiable as of 2026-10-10 from the available web sources, because the results do not provide a chain-level collateral composition or on-chain exposure breakdown for Ethereum. The only directly relevant risk statement available is that Spark Savings’ stated risks include depeg risk for USDS-backed savings products, and V2 USD vaults are described as fully backed by USDS. What can be said from the sources is narrower: Spark Savings V2 USD vaults (spUSDC, spUSDT, spPYUSD) are described as fully backed by USDS, and Spark states that all USD Savings Vaults share the same underlying risk profile because of that backing.

Spark also says Savings uses a liquidity buffer of up to $10 million for typical withdrawals, while excess liquidity is deployed through the Spark Liquidity Layer into yield strategies. However, the sources do not identify which collateral is the largest on Ethereum, nor do they quantify the Ethereum-only share of assets, so the stress loss from a 20% depeg cannot be computed reliably here. Any numeric estimate would be speculative.

Scope caveat. “Spark Savings V2 Onchain” does not identify a single Ethereum vault: V2 includes spUSDC, spUSDT, spETH and other products with different allocation profiles. The current top counterparty, exposure amount, and concentration are Not verifiable as of October 10, 2026 because Dune/on-chain verification is unavailable and the public documentation does not provide a current per-vault exposure table. Stress scenario — largest allocated counterparty becomes insolvent 1. Initial loss path: The counterparty/strategy fails to return assets. The Spark Liquidity Layer (SLL) records a shortfall in the affected allocation; recovery value may depend on collateral liquidation, legal recovery, or residual asset value.

The vault’s economic backing is impaired even if the ERC-4626 share token continues trading. 2. First-loss absorption: Losses are intended to hit the relevant Prime’s internal junior risk capital first, followed by any Prime-external junior capital allocated to that exposure. These are the primary compensation layers and are not claims on individual depositors.

3. Senior backstops: After junior capital is exhausted, the documented waterfall proceeds to external senior risk capital through srUSDS, if deployed; then Sky’s Surplus Buffer; then the Aggregate Surplus Buffer. The documentation describes srUSDS as planned, so its live availability is Not verifiable as of October 10, 2026. 4. Residual loss: If those resources are insufficient, Sky may mint SKY for recapitalization.

Only after junior capital and token backstops are exhausted does the documentation describe residual losses being socialized across USDS holders, including USDS-backed Spark Savings vaults. This is a protocol loss-absorption design, not a guaranteed depositor-compensation scheme. Smart-contract impact path: ERC-4626 vault shares represent claims on the vault’s assets; redemptions burn shares and transfer underlying assets. Spark can place affected vaults into recovery mode and temporarily halt withdrawals.

Normal withdrawals use the vault; larger withdrawals may use the Savings Vault Intents contract, which calls redeem() after liquidity is prepared. Insolvency therefore flows through asset impairment, controller/governance reallocation and potentially reduced redemption value or delayed withdrawals—not an automatic insurer payment.

For the DAO/owners committing fraud stress scenario, I did not find evidence of an actual fraud event for Spark Savings V2 on Ethereum; the available materials instead show a governance-controlled design with explicit administrative power and centralized roles, which creates *fraud-enabling governance risk* rather than a verified fraud incident. The strongest documented concern is that the DEFAULT_ADMIN_ROLE is fully trusted and can, in the worst case, DoS the system or whitelist a malicious vault, while the permissioned TAKER_ROLE is granted to the Spark Liquidity Layer to move idle liquidity into strategies. Audit material also states that the relayer is semi-trusted and that blocked assets or recipients can prevent fulfillment, underscoring operational and counterparty control risk.

For this stress case, the practical loss mechanism would be governance abuse: malicious owners/DAO actors could redirect vault flow, whitelist hostile strategies, or otherwise misuse privileged roles to expropriate or freeze funds. That risk is not the same as proven fraud, but it is a credible severe-loss scenario because the system relies on trusted governance and permissioned execution paths. Spark’s docs also frame Savings V2 as using the Spark Liquidity Layer and permissioned roles, which confirms the presence of centralized control points. Assessment:

• Likelihood of catastrophic loss if privileged governance is compromised: high, because trusted roles can alter execution paths.
• Evidence of committed fraud by DAO/owners: not found in the provided sources.
• Evidence of centralized control risk: documented. Not verifiable as of 2026-10-10: whether any DAO/owner has actually committed fraud against Spark Savings V2 users.

For Spark Savings V2 on Ethereum, a stress case with a negative 30-day primary yield source means the protocol’s main yield engine is underperforming or generating losses over the trailing month. Spark’s own documentation says V2 yield is produced through the Spark Liquidity Layer and governed rates, not inside the vault itself, so a negative source would pressure the vault’s ability to maintain the quoted APY and could compress depositor returns. The key risk is rate transmission: if the underlying yield strategies go negative, the vault can still accrue according to its accumulator mechanics, but only if the external rate setting remains supported by governance and reserve economics.

If the negative 30-day condition persists, the likely outcomes are APY reduction, slower share-value growth, and potential reliance on governance to reprice the vault rate or alter allocations. For this specific Ethereum deployment, I cannot verify on-chain exposure, realized 30-day source yield, buffer usage, or loss coverage in this run. Not verifiable as of 2026-10-10. The protocol docs confirm the architecture, but they do not provide a verifiable stress-loss absorption rule for a negative 30-day primary source on Ethereum.

Scope: Ethereum Spark Savings V2 (spUSDC/spUSDT/spETH). Control map: contracts and material parameters are controlled through Spark’s SubDAO proxy (0x3300…6af8c4) and Sky/Spark governance. The vault grants a whitelisted TAKER path to the Spark ALM Controller, which deploys liquidity; the ALM Operations Safe sets operational parameters. Spark Foundation retains ownership of strategy/code determining capital allocation, while a development company executes it.

The frontend/interface is separate from the non-custodial protocol and is operated through the Sky/Spark ecosystem; its operator does not directly control contract state. Governance: real but nested, not fully autonomous. SPK holders vote via Snapshot; proposals require 1% of SPK supply (100m), Spark Risk Council and Operational Facilitator review, then >50% approval. Passed proposals update the Sky Atlas and are implemented through Spark/Sky spells.

Therefore dao_governance=true, but Sky governance and designated councils retain material gating power. Voting concentration and top SPK holders: Not verifiable as of October 10, 2026 (Dune unavailable; no on-chain query ID/execution ID can be supplied). Timelock: Sky executive changes affecting Spark are subject to the GSM Pause Delay, currently 48 hours. This is a governance execution delay, not necessarily a dedicated vault-local timelock. Funds/admin risk: admin_can_drain=true under the requested definition: the operational ALM/taker/controller path can draw vault liquidity and move assets into approved strategies without a new governance vote.

Rate limits, whitelisting and freezer controls constrain this authority; it is not evidence of unrestricted arbitrary withdrawal by one EOA. A relayer can fulfill withdrawal intents but cannot redirect the user’s redemption assets according to the reviewed design. Emergency bypass: Not verifiable as of October 10, 2026 for a governance-bypassing path specific to these Ethereum V2 vaults. Freeze/DoS controls are documented, but a bypass enabling value-changing execution was not established.

Legal entities disclosed: SPK Company Ltd (British Virgin Islands) and Spark Foundation (Cayman Islands). Registration number, directors and applicable Spark frontend ToS details: Not verifiable as of October 10, 2026.

Scope/name-collision check: this is Spark Finance/Spark Savings V2 on Ethereum (spUSDC/spUSDT), not Lightspark’s unrelated “Spark” Bitcoin project. No Dune/on-chain verification was available in this run. Legal structure: Spark’s MiCA white paper identifies SPK Company Ltd.

(company limited by shares, registered/head office in the British Virgin Islands) as token offeror, with Spark Foundation in the Cayman Islands as parent. It describes Spark as a Sky ecosystem “subDAO,” but does not clearly identify a separate legal operating entity for Spark Savings V2. ToS/restrictions: I found no clearly applicable, protocol-level Savings V2 Terms of Service.

The app is presented as non-custodial and users interact with ERC-4626 vault contracts. Campaigns may restrict countries, and Spark may block wallet access based on prohibited-conduct screening. KYC/AML/data protection: The documented control is wallet/address screening by TRM Labs; Spark states the wallet address is shared, while IP metadata is not shared with TRM.

This is not evidence of customer KYC or a full AML program. A comprehensive privacy notice, controller identity, retention policy, and data-subject rights for Savings users were not verifiable as of October 10, 2026. Classification: The MiCA white paper states SPK is not intended to be a financial instrument, deposit, insurance product, or other regulated financial product under EU law, and is not approved by an EU competent authority.

That statement concerns SPK, not the legal classification of Spark Savings vault shares or yield products. The vault tokens are transferable ERC-4626 instruments, creating potential securities, collective-investment, lending, payments, and consumer-protection questions by jurisdiction; no regulator classification was found. Enforcement/courts/sanctions: No regulator action, court case, or sanctions designation against Spark Foundation, SPK Company Ltd., or Spark Savings itself was identified in the reviewed sources.

This is not a clearance opinion; broader searches and entity-name ambiguity limit certainty. Legal risk remains materially higher than the foundation/company structure suggests because governance, smart-contract control, liquidity allocation, and user access are distributed across Sky governance, interfaces, service providers, and contracts. Active enforcement: false (no identified action; not a guarantee) Sanctioned: false (no identified designation; interface screening is not a sanctions designation)

No exact GLEIF LEI record for 'Spark Foundation', 'Spark Savings V2 Onchain'. OFAC SDN screening of 'Spark Foundation', 'Spark Savings V2 Onchain': no match. SEC litigation and administrative release feeds: no mention.

Spark Savings V2 on Ethereum does not appear to issue its own stablecoin; the relevant vaults are denominated in USDC/USDT/PYUSD/ETH, while the protocol’s own docs describe depeg risk specifically for USDS/sUSDS, not for Spark-issued stablecoin assets. For the selected Ethereum V2 savings vaults, a stablecoin depeg event is not verifiable as of 2026-10-10, because no on-chain verification is available in this run and the gathered web sources do not provide a documented depeg history for the vault’s underlying deposit stablecoin(s).

Spark Savings V2 on Ethereum combines ERC-4626 vault, liquidity-allocation, governance, and USDS-backed settlement risks. Audits identify generally high code security, but also note liquidity-dependent behavior and reliance on conservative configuration. On-chain TVL, allocation concentration, and reserve coverage are Not verifiable as of October 10, 2026 because Dune was unavailable.

Spark Savings V2’s top strengths are: (1) conservative risk management, because it minimizes exposure to riskier collateral and uses a disciplined allocation framework; (2) strong liquidity and withdrawal support, with instant liquidity buffers and asynchronous intents for larger withdrawals; (3) institutional-scale capacity, including support for large transactions and 24/7 withdrawals at individual depositor size; (4) broader asset coverage and composability, expanding from USDC-only to assets like USDT, PYUSD, and ETH while remaining ERC-4626-based; and (5) diversified yield sourcing via the Spark Liquidity Layer, which allocates across DeFi and RWA sources to target better risk-adjusted returns than traditional lending. A few supporting details matter for an institutional view: Spark says Savings V2 maintains up to a $10 million liquidity buffer for typical atomic redemptions, and larger withdrawals are usually fulfilled within minutes through the liquidity intents mechanism. It also describes SLL as the system that pulls idle liquidity, deploys it into yield-bearing strategies, and returns proceeds to the vault, which supports continuous share-value accrual.

If you want, I can turn this into a risk-factor balanced strengths/weaknesses memo for Spark Savings V2 on Ethereum.