SparkLend

SparkLend is a decentralized, non‑custodial liquidity / money market protocol built within the Spark / Sky (ex‑MakerDAO) ecosystem and deployed on Ethereum mainnet only as of the latest data. Because on‑chain tools are unavailable this run, all on‑chain facts are Not verifiable as of 2026‑08‑26. ### Protocol identification

• Name / brand: SparkLend (product within the broader Spark / Sky protocol stack).
• Category: DeFi lending / money market protocol (stablecoin‑focused, but supports majors like ETH, WBTC, USDC).
• Website / app: SparkLend is surfaced via the Spark front‑end at spark.fi and its Borrow section.
• Docs: Technical and product documentation under Spark Documentation Portal, SparkLend product page.
• Chains: Documentation and analytics sources state Ethereum only for SparkLend deployments.
• Launch date: Multiple analytics / research pieces state Spark launched on Ethereum mainnet in May 2023; SparkLend is part of this initial deployment.
• Native token: SparkLend itself has no dedicated native token; governance and risk parameters are controlled by Sky/MakerDAO governance via MKR executive votes. ### Main contracts (high‑level only)
• Public GitHub shows repository sparkdotfi/sparklend-v1-core, explicitly marked as forked from aave/aave-v3-core, indicating standard Aave v3 money‑market architecture (Pool, Configurator, Tokens, etc.).
• Specific Ethereum contract addresses and explorer verification status are Not verifiable as of 2026‑08‑26 under current constraints. ### Fork lineage and modifications
• Multiple independent sources agree SparkLend is an Aave v3 fork operated under Sky governance.
• Key changes vs upstream Aave v3 (per independent research, not on‑chain verified):
• Collateral set tightly curated around highly liquid assets and Maker/Sky ecosystem assets (DAI, sDAI, USDS, sUSDS, ETH, stETH, WBTC, etc.).
• Rate model: governance‑defined / fixed‑style lending rates aligned to DAI/Sky Savings Rate and stETH yields, rather than purely utilization curves.
• Liquidity sourcing: direct credit lines (D3M‑like) and the Spark Liquidity Layer (SLL) supplying stablecoin liquidity from Sky treasury/vaults.
• Risk management: multi‑oracle pricing (RedStone, Chainlink, Chronicle), narrow collateral, strict rate limits, and first‑loss capital. ### Audits / malicious‑modification history
• Public info confirms open‑source Aave v3‑fork repo but does not clearly enumerate dedicated third‑party audits of SparkLend’s modifications; audit status for the forked codebase is therefore Not verifiable as of 2026‑08‑26.
• No independent reports of malicious modifications or exploit‑driven forks specific to SparkLend were found; however, absence of evidence is not proof of safety. Not verifiable as of 2026‑08‑26.

SparkLend appears to have a real, functioning product portal rather than a pure landing page: the docs point users to app.spark.fi, describe active borrowing/supplying flows, and note supported networks including Ethereum, while the main site advertises direct access to SparkLend from the app. The documentation is fairly mature, with dedicated product, getting-started, troubleshooting, E-Mode, audits, oracle, caps, and developer pages, which is a strong sign of a live operational UX rather than a template shell. Live deposits/withdrawals are not independently verifiable in this run, so that remains "Not verifiable as of 2026-08-26". The same applies to any claim about exact live usage, broken links at scale, or whether specific metrics displayed on the site are real versus promotional. On open API availability, there is evidence of programmatic access: the developer ecosystem references Spark-related API endpoints for markets, deposits, debts, and health checks, but that source is a third-party GitHub project, so it supports API existence only indirectly. The official docs also frame Spark as accessible through the app, API, or directly via smart contracts, but that is still best treated as an unverified product claim until independently confirmed. Overall: mature portal and docs, likely live app functionality, but open API status is only partially supported and not fully verifiable here.

The audit reported one high-severity issue: the weETH oracle could be manipulated upwards by burning eETH, which could potentially drain the protocol. Spark acknowledged the issue but deemed it not a practical risk under current weETH collateral parameters on SparkLend Mainnet and because borrowing is disabled. The report also states a stable-debt calculation issue was corrected, with totalDebt removed and totalVariableDebt used directly.

SparkLend-related core audits were performed on the SparkLend Smart Contract Audit and related Spark-specific core changes built on the Aave v3 codebase. Spark docs state SparkLend is based on Aave v3, with notable custom changes including a DAI market interest-rate strategy, a custom oracle for Savings DAI, and a RewardsController setup difference. The docs also state the Spark-specific core audits cover protocol-level core updates and deployment changes.

The audit concluded that after the intermediate report all issues had been resolved, and overall the codebase provides a good/good level of security. The report frames the review as time-boxed and not a guarantee against all vulnerabilities.

Cap Automator was audited separately. The report says the most critical subjects were functional correctness, manipulation resiliency, and integration into SparkLend. It identified an issue where setting caps to zero was not restricted, which could bypass cooldown and risk lifting the cap; after the intermediate report, all identified issues were addressed or acknowledged.

SparkLend has an active bug bounty program managed by Immunefi. Immunefi lists it as live since 01 November 2023, with the page last updated on 13 August 2026. The program’s maximum bounty is $5,000,000, and the reward structure states 10% of the funds directly affected up to that cap. On scope, Immunefi’s program resources page shows the program as triaged by Immunefi and the BountyIndex snapshot reports 359 in-scope items as of an update on 2026-08-08. Spark’s own documentation also says it has an active bug bounty program with rewards of up to $5,000,000. For results, the publicly available search results do not provide a verifiable tally of disclosed payouts, valid submissions, or exploit recoveries tied specifically to this bug bounty. Therefore, the results are: Not verifiable as of 2026-08-26.

SparkLend’s main external dependency is its oracle stack. Its docs say markets use Chronicle, Chainlink, and RedStone, with a three-oracle median in normal conditions, a redundant/average fallback if one feed is missing, and a Uniswap TWAP fallback if all three fail; that reduces single-oracle failure risk but creates residual dependence on third-party feed integrity and Uniswap liquidity/price quality. The clearest counterparty/manipulation concern is around LST/LRT pricing. ChainSecurity noted that wstETH, rETH, and weETH oracles can misprice during an LST/LRT depeg, and that the weETH oracle can be manipulated upward by burning eETH; it judged the intended configuration to be safe, but still identified a path where manipulation could potentially drain the protocol. Spark’s own docs also say a peg-ratio oracle is used for correlated collateral such as wstETH, rETH, weETH, cbBTC, WBTC, and LBTC, and a kill switch halts new borrowing if deviation thresholds are breached. Exposure to stablecoins is material because SparkLend explicitly supports stablecoin-correlated borrowing and collateral such as USDC, USDT, sDAI, sUSDS, and PYUSD. That means depeg or issuer distress in any listed stablecoin would transmit directly into collateral valuation, liquidation, and borrow-side risk. The docs do not identify SparkLend as relying on a single custodian, bridge, or centralized exchange/MM for core lending on Ethereum; those dependencies are not verifiable as of 2026-08-27. Spark also appears to depend on external asset protocols for some collateral types and integrations, including Lido-style LSTs and newer restaking/LRT assets referenced in the audit and docs, so failure risk extends to upstream staking, restaking, or wrapped-asset systems rather than only Spark itself. Any RWA issuer/SPV exposure is not verifiable as of 2026-08-27. In practice, the main downside scenarios are oracle failure, oracle manipulation, LST/LRT depeg, stablecoin depeg, or upstream asset protocol insolvency/collateral impairment.

SparkLend is organized as a non-custodial lending protocol on Ethereum: users interact with permissionless smart contracts, keep control of their assets in their own wallets, and the protocol itself does not provide wallet infrastructure or custody services. In practice, that means lenders and borrowers hold the relevant keys themselves; if private keys or wallet access are lost, access to the tokens or positions can be permanently lost. For the core SparkLend market, collateral supplied to reserves stays in the reserve and is not redeployed into external strategies, while cross-module transfers are controlled by smart-contract rate limits. This reduces operational discretion: custody is not centralized with Spark, but enforced by contract logic and the user’s wallet security. There is one important institutional exception: Spark has announced an Anchorage Digital integration for institutional crypto-backed lending, where collateral can remain in Anchorage Digital Bank’s custody while institutions access Spark liquidity through Atlas collateral-management workflows. That arrangement is separate from the normal retail/non-custodial SparkLend user flow and indicates that, for some institutional setups, custody may sit with Anchorage rather than the borrower. So, for Ethereum SparkLend, the default custody model is user-self-custody via smart contracts, with select institutional workflows using an external qualified custodian such as Anchorage.

SparkLend’s response, as described in the sources, was pre-emptive risk reduction: it halted new rsETH supply and removed rsETH from approved collateral before the April 2026 exploit, which is presented as the reason it had no direct losses.

The only clearly described incident in the source set is the April 18, 2026 Kelp DAO / LayerZero bridge compromise affecting rsETH, but the loss was borne by Aave rather than SparkLend; several reports say the attack minted about 116,500 unbacked rsETH and led to roughly $124 million to $230 million in bad debt on Aave, while SparkLend saw inflows instead.

SparkLend is non-custodial: users control their own assets and wallet access, and the protocol cannot restore funds lost through private-key loss. In practice, key management is organized at the user/wallet level, not by a centralized protocol operator. Spark’s docs say users must use a self-custodied wallet, securely store private keys and recovery phrases, and pay Ethereum gas from that wallet. The terms of service also state that users are solely responsible for safeguarding the private key for the address used with the Spark smart contracts, and that the site operator will not restore property lost due to private-key loss. For Ethereum specifically, SparkLend is described as operating on Ethereum mainnet as a permissionless money market, so the relevant key-management model is standard Ethereum wallet custody rather than a protocol-managed key vault. I did not find evidence of any SparkLend-administered key custody, multisig for end-user funds, or centralized key recovery mechanism in the provided sources; that is Not verifiable as of 2026-08-26.

SparkLend on Ethereum is an upgradeable proxy-based system centered on the PoolAddressesProvider, which acts as the registry and proxy admin for core protocol contracts; the docs describe it as the single source of truth for Pool, PoolConfigurator, Price Oracle, and ACL Manager addresses, and state that on Ethereum mainnet it is owned by Aave Governance. Etherscan’s verified SparkLend proxy page explicitly describes the proxy as combining upgradeability with administrative authorization, and a separate verified contract page shows SparkLend deployment constants including a Proxy Admin address and a Transparent Proxy Factory, consistent with a proxy/implementation architecture. The deployment verification PDF lists multiple Spark Lend proxy addresses, indicating a multi-contract deployment rather than a single immutable contract set. Admin / control surface: PoolAddressesProvider can update protocol addresses and configurations; PoolConfigurator, ACL Manager, and Price Oracle are all governed through that registry. The documentation does not, in the retrieved material, prove whether emergency/pause/withdrawal/upgrade/fee/oracle strategy roles are renounced or permanently disabled; therefore that status is Not verifiable as of 2026-08-26. Exit / freeze risk: Because SparkLend is a lending market with proxy-controlled configuration, a compromised governance/admin key could plausibly upgrade implementations, change critical addresses, or alter risk parameters; however, whether users can *always* exit depends on market liquidity and protocol state, and the retrieved sources do not verify a hard non-custodial escape guarantee, so this is Not verifiable as of 2026-08-26. Rug/freeze assessment: The architecture is not an immutable, ownerless design; it has meaningful admin/governance control by construction. That means the principal worst case is governance or proxy-admin compromise leading to hostile upgrades or parameter changes that could freeze markets or redirect control, although the sources reviewed do not show evidence of such an event. Architecture map: Aave Governance / governance-controlled admin → PoolAddressesProvider (registry + proxy admin) → Pool / PoolConfigurator / Oracle / ACL Manager (upgradeable components) → Transparent proxies + implementations Diagram: Governance/Admin → PoolAddressesProvider → {Pool, PoolConfigurator, Oracle, ACL Manager} → User deposits/borrows Key missing items remain Not verifiable as of 2026-08-26: exact current on-chain timelock delay, role renunciations, and a Dune-based proxy-admin event classification, because on-chain verification was unavailable in this run.

SparkLend is presented by third-party coverage as a DeFi protocol developed within the Sky/MakerDAO ecosystem, with Rune Christensen named as founder and Lucas Manuel as a technology lead; however, the web results do not verify a dedicated legal entity, office, or onshore/offshore structure for SparkLend itself. The strongest identity signal in the results is that Spark appears to be a product line of Sky/MakerDAO rather than a standalone anonymous team, which points to a public, identifiable leadership model rather than an anon core. Credibility-wise, Rune Christensen is widely known for founding MakerDAO and creating DAI, but that reputation belongs to the broader Sky/MakerDAO history rather than being independently documented here for SparkLend-specific operations. The available results also suggest the team overlaps with former Compound engineers and Celestia advisors, but this is not independently substantiated in the results beyond promotional or secondary coverage. Reality check: I found no verifiable evidence of a standalone SparkLend office, incorporated headquarters, or separate operational footprint; the available material mostly describes Spark as software developed by the Sky/MakerDAO team and marketed through web properties and social accounts. That means the “real business vs. web front” question is only partly answerable from these results: there is clear evidence of named leadership and a recognizable parent ecosystem, but not enough independent evidence of a separate physical organization for SparkLend itself. Not verifiable as of 2026-08-26.

SparkLend’s reputation is generally positive among DeFi risk reviewers: it is described as a conservative, permissionless lending market with multiple audits, a public bug bounty, and a long operating history without any widely reported major breach or exploit in the sources reviewed. Independent commentary also characterizes the team as verifiable and the project as “legitimately constructed,” while noting that the main risks are structural rather than fraudulent.

SparkLend is an overcollateralized money market for DAI and ETH-centric assets, designed as the lending arm of Maker’s Spark ecosystem, with yields primarily coming from borrower interest and Maker-linked DAI facilities, not speculative farming. Strategy & assets in/out Users supply DAI, ETH and selected LSDs (e.g., stETH, wstETH) and can borrow mainly DAI and ETH against them. The design closely mirrors Aave v3 but is governed by Maker/Spark governance. Yield source: organic vs subsidized Supplier APY is mostly organic: interest paid by borrowers, set by interest rate curves per market. A major additional source is the DAI Savings Rate (DSR) and Maker yield on DAI held and deployed by Spark/Maker (via sDAI and associated vaults), making DAI yields partly protocol-routed Maker income rather than external incentives. There are no large ongoing liquidity mining programs; any MKR/other rewards are minor or time-bound—overall low subsidy profile. Risk profile: market-neutral vs directional, leverage For simple DAI lenders, exposure is close to market-neutral credit risk of borrowers plus Maker/Smart contract risk. ETH/LSD suppliers remain directionally long ETH, with additional leverage possible via looping (e.g., deposit wstETH, borrow DAI, buy more wstETH). Spark also connects to Maker’s DAI-centric leveraged strategies (e.g., vault-backed lines) at protocol level, but retail leverage is via standard borrow/loop mechanics. Lock-ups & withdrawals SparkLend is fully liquid under normal conditions: deposits can be withdrawn anytime subject to pool liquidity. No fixed lock-ups, but utilization spikes or risk parameters (e.g., caps) can delay full withdrawal. Fees, gates, limits Users pay borrow interest only; no explicit deposit or withdrawal fees. Risk is controlled via supply and borrow caps, collateral factors, liquidation thresholds, and interest rate slopes, adjustable by governance. Liquidations follow standard DeFi patterns with discounts for liquidators and penalties borne by undercollateralized borrowers. Protocol revenue Economic surplus accrues to Maker/Spark via the spread between what Spark earns on DAI (DSR, vault yield) and what it pays to depositors, plus any interest margin from lending operations. Borrow interest is split between deposit APY and protocol reserve. Collateral & TVL / APY data Detailed TVL by asset, protocol revenue, and historical APY volatility are Not verifiable as of 2026-08-27 (no Dune; DeFiLlama/other aggregators cannot be cross-checked on-chain under current constraints). APY history and sustainability must therefore be treated as aggregator-only estimates, not on-chain verified figures.

SparkLend’s reserve/tresury picture is only partially verifiable from the available web sources; the protocol docs confirm that SparkLend has reserve accounting at the pool level (including an accrued-to-treasury field, reserve factors, caps, and reserve configuration), but the actual treasury size, custody addresses, and on-chain balances are not verifiable as of 2026-08-27 without an on-chain query. What can be stated confidently is that SparkLend runs on Ethereum and its reserve configuration includes multiple assets such as DAI, USDC, WBTC, WETH, sDAI, and wstETH; the docs also show that reserve factor varies by asset, with examples including 10% on DAI/USDC/WBTC and 5% on WETH in the deployment verification appendix. The pool contract exposes reserve-related state such as accrued-to-treasury, unbacked, and reserve caps, which indicates treasury accrual is tracked inside the lending pool rather than described here as a separate public treasury wallet. For composition, the available sources point to SparkLend’s broader reserve base being concentrated in Ethereum-market lending markets and supported by Sky ecosystem liquidity, but those claims come from analytics or media rather than raw on-chain verification. Because Dune is unavailable in this run, the reserve size, treasury address set, custody structure, and exact token balances cannot be confirmed on-chain. No independent attestation for a SparkLend treasury wallet or reserve proof was verifiable from the provided sources. The only directly relevant attestations found are deployment verification and pool-interface documentation, which validate contract mechanics but do not provide a current treasury balance sheet.

SparkLend does have a native token: SPK on Ethereum. The sources found indicate SPK is the governance/staking token for the Spark ecosystem, but key on-chain tokenomics items are Not verifiable as of 2026-08-28 without direct chain verification. What is supported by the web sources: SPK is described as Spark’s native token and governance core, with use in governance, staking, and ecosystem farming. A governance proposal also states SPK incentives for SparkLend pre-farming and references Spark–Aave revenue share payments. The tokenomics sources found also describe a 10 billion total supply and a 10-year emission roadmap, with 6.5 billion for Genesis farming and 3.5 billion for the Spark Foundation, and emissions starting at 1.625 billion/year before tapering to about 203 million/year by year seven. However, the actual token contract address, total vs. circulating supply, market cap, FDV, unlock schedule execution, top-holder concentration, insider wallets, mint/blacklist/fee-switch controls, and DEX liquidity depth are Not verifiable as of 2026-08-28 from the available evidence. Tokenomics items that remain unverified: contract address; current circulating supply; market cap and FDV; whether revenue share is ongoing or only episodic; buybacks, burns, or staking reward mechanics; detailed allocations to team, investors, treasury, and community; whether announced unlocks actually happened on-chain; and holder concentration/insider wallet analysis. The protocol sources suggest SparkLend governance is tied to the Sky/MKR → SKY transition and that SPK governance is being used for Spark-level decisions, but control-path details for mint/blacklist/fee-switch functions are not verifiable as of 2026-08-28. The DEX listings found show SPK pairs on Ethereum/Uniswap, but liquidity depth is not verifiable as of 2026-08-28.

SparkLend is a conservative lending market on Ethereum with multi-oracle pricing, borrow caps, and liquidation rules designed to limit insolvency risk under stress. For the specific scenario “bitcoin price falls below $10,000,” the key stress channel is WBTC/cbBTC collateral devaluation: if Bitcoin’s market price collapses, positions backed by BTC-linked collateral would see their health factors fall and become liquidatable once they cross the protocol’s threshold; SparkLend states that positions with health factor below 1 are liquidatable, with up to 50% of debt covered between 1.00 and 0.95 and up to 100% at or below 0.95. SparkLend also says hard-pegged / exchange-rate collateral such as WBTC and cbBTC is protected by a peg-ratio oracle and a killswitch that halts new borrowing if deviation thresholds are breached, which is specifically meant to reduce the risk of distressed collateral being accepted at stale value. What cannot be verified from the provided sources is the actual protocol loss, bad debt, or TVL impact from a BTC crash to $10,000; those figures are not available in the search results, and on-chain verification is not available in this run. Not verifiable as of 2026-08-26. Operationally, the likely outcome in this scenario is more liquidations, tighter BTC-collateral capacity, and potentially reduced borrowable liquidity for users using BTC as collateral, but the magnitude cannot be quantified from the available evidence.

SparkLend on Ethereum has explicit depeg-aware controls for hardcoded/exchange-rate collateral such as wstETH and rETH: a peg-ratio oracle monitors market price versus underlying, and if the deviation breaches the asset-specific threshold, a killswitch stops new borrowing against that collateral. In a 20% collateral depeg stress, the immediate effect is therefore parameter-dependent and not fully quantifiable from the public docs alone: whether the market hits the threshold, and by how much liquidation conditions worsen, are asset-specific and threshold-specific. The public docs do confirm the liquidation mechanics that would govern the stress outcome. If a position’s Health Factor falls below 1 but stays above 0.95, up to 50% of its debt can be repaid in a liquidation call; at or below 0.95, up to 100% can be covered. Spark also states that borrowing power for an asset can be reduced to 0% without forcing existing borrowers out first, which gives governance a lever to de-risk a collateral type during stress. For a precise “largest collateral depegs 20%” loss estimate on SparkLend Ethereum, the key missing inputs are the specific asset, the active threshold, the distribution of borrower health factors, and the protocol’s current exposure to that collateral. Not verifiable as of 2026-08-26.

For SparkLend on Ethereum, the main insolvency stress path is a borrower or collateral venue failing to repay while a position’s health factor falls below 1, triggering liquidation through liquidationCall(); if health factor is between 0.95 and 1, up to 50% of debt can be covered, and at or below 0.95, up to 100% can be covered. In that path, liquidators absorb execution risk first by buying the collateral at the liquidation bonus, while depositors/lenders absorb residual loss only if liquidation cannot fully cover the debt; Spark’s docs also state that a single depositor or single adverse event cannot drain the protocol in one block and that rate limits cap capital at risk on any route per unit time. If the top counterparty is insolvent in the sense of a major upstream venue or ecosystem counterparty failing, the publicly available materials do not provide a protocol-specific, on-chain loss waterfall for SparkLend itself; the exact compensation path is Not verifiable as of 2026-08-26. The most relevant documented mechanism is still the liquidation path: collateral is transferred based on debt covered, oracle prices, and liquidation bonus, and the debt is repaid from the liquidator’s capital rather than from protocol treasury by default. The impact path through smart contracts is: oracle price update  health factor deterioration  account becomes liquidatable  liquidationCall() executes  collateral is seized and debt is repaid  bad debt is limited to any uncovered remainder. Chainlink’s general description of DeFi lending matches this structure: smart contracts monitor collateralization in real time and automatically sell collateral to protect lenders from bad debt. I cannot verify, from the provided sources, any Ethereum-specific SparkLend backstop fund, treasury compensation rule, or insolvency waterfall beyond liquidation and rate-limit controls; any such claim would be Not verifiable as of 2026-08-26.

For SparkLend on Ethereum, I found no evidence of committed fraud by the DAO or owners in the provided sources. The most directly relevant independent write-up explicitly says it found “no evidence of fraud, manipulation, or rug-pull architecture” and characterizes the risk as systemic rather than fraudulent . What *is* verifiable from the sources is a risk-management posture, not a fraud event: SparkLend’s rate limits, caps, oracle kill switch, and earlier reduction of rsETH exposure are described as the reason it avoided losses during the KelpDAO exploit . Another source notes a code-audit issue where the weETH oracle could be manipulated upward by burning eETH, which could potentially drain the protocol if exploited, but that is a technical vulnerability, not evidence of owner or DAO fraud . So for the stress scenario “committed fraud by the DAO or owners,” the current assessment is: not evidenced. On the material provided, the appropriate answer is Not verifiable as of 2026-08-26 for any fraud committed by SparkLend’s DAO or owners, and the available evidence points instead to non-fraud operational and governance risk .

For SparkLend on Ethereum, the primary yield source is not verifiable as negative over the last 30 days from the available non-onchain sources. The strongest available evidence says SparkLend is the default/highest-priority venue for USDT in Spark’s risk architecture, and Spark’s ETH vault risk is mainly exposed to SparkLend ETH lending market activity. What can be said with confidence is that Spark’s yield architecture is multi-source, not a single fixed stream: Spark Savings ETH is composed mainly of ETH deposited in SparkLend’s ETH market plus idle reserves, and Spark’s SSR is described as funded by Sky protocol revenue, including Spark borrow interest and other collateral income. That means a negative 30-day result for one yield leg would not by itself prove the protocol’s overall yield source is negative. There is also no reliable, independent 30-day return series in the provided sources for SparkLend Ethereum that would let me verify a negative primary yield source. The protocol docs and related analyses describe risk controls, collateral scope, and venue hierarchy, but they do not provide a 30-day negative yield measurement for the primary source. Bottom line: Not verifiable as of 2026-08-26 whether SparkLend’s primary yield source on Ethereum was negative over the past 30 days.

SparkLend is not credibly DAO-controlled in the fully sovereign sense; the governance path that can change protocol parameters and trigger Spark spells runs through Maker/Sky governance, with GSM Pause Delay timelocks on execution, while certain emergency/security actions use dedicated multisigs approved by that governance process. The clearest on-record process is: community/forum discussion, Maker/Sky executive voting, then delayed on-chain execution via a spell after the GSM Pause Delay; one governance note states the delay was 16 hours at that time. Control appears split as follows: protocol parameters and spells are governed through Maker/Sky executive votes; emergency security access can be delegated to a SparkLend External Security Access multisig for the freezer module; and the protocol’s own documentation trail shows repeated dependency on Maker governance for execution rather than a self-contained Spark DAO. That makes the DAO partly real but materially constrained/symbolic relative to a fully independent protocol DAO, because the decisive authority sits upstream in Maker/Sky governance. The emergency multisig referenced publicly is 0x44efFc473e81632B12486866AA1678edbb7BEeC3 for SparkLend External Security Access to the freezer module. A separate source also lists a SparkFoundationMultisig and other named multisigs, but signer set, threshold, and signer independence are Not verifiable as of 2026-08-27 from the gathered sources. For voting concentration, top-holder distribution, and concentration of governance power, Not verifiable as of 2026-08-27 because the required on-chain analysis was not available in this run. On the company/legal side, the terms page says Spark Lending is operated by RWHS, LLC and gives a U.S. address in Pearland, Texas, USA; jurisdiction, registration number, directors, and whether that entity directly controls SparkLend contracts/frontend/funds beyond the terms page are Not verifiable as of 2026-08-27. No verified source in this run established a single entity that independently controls all dev, contracts, frontend, and funds; the best-supported reading is Maker/Sky governance for protocol changes, multisigs for emergency actions, and RWHS, LLC for the branded service terms.

SparkLend is part of the Spark Protocol/MakerDAO ecosystem, with its web front end operated by Spark.fi Inc. and related entities in offshore jurisdictions; the on-chain protocol itself remains a permissionless, non‑custodial lending market on Ethereum. ### Legal entity & governing law

• The Spark.fi site is operated by *Spark.fi Inc.*; its Terms of Use are governed by Cayman Islands law.
• Separate airdrop terms reference a different entity structure, with disputes governed by British Virgin Islands law and arbitrated under BVI IAC rules in Tortola.
• These fragmented documents point to a multi‑entity, offshore web‑frontend/operator model typical for DeFi front ends, while the protocol contracts are presented as decentralized infrastructure. ### User restrictions, geoblocking & VPN
• Media reporting notes that Spark’s front end displays: “Accessing this website via VPN is not allowed,” and terms indicate opposition to using VPNs to evade geographical restrictions, particularly targeting U.S.-based users.
• This suggests front‑end level geofencing and behavioral restrictions, likely aimed at reducing U.S. securities/commodities enforcement risk, even though the underlying protocol is still accessible at contract level. ### KYC/AML & classification
• SparkLend is documented as a fork of Aave v3 operated by the MakerDAO / Sky ecosystem and marketed as non‑custodial DeFi lending.
• Public reviews characterize Spark Protocol as unlicensed, decentralized, relying on “blockchain-specific compliance” rather than traditional regulatory licenses.
• As of today, there is no evidence of protocol-level KYC/AML for SparkLend smart contracts; users interact via Ethereum addresses. This matches broader analysis that DeFi lending protocols typically conduct no KYC at the protocol layer.
• Some off‑chain integrations (e.g., access to USDS via centralized exchanges) may involve exchange‑level KYC/AML, but that is handled by those custodians, not SparkLend itself. ### Enforcement, warnings, sanctions, and court cases
• No independent records of regulatory enforcement actions, formal warnings, sanctions listings, or court cases specifically against SparkLend, Spark.fi Inc., or Spark Protocol could be identified.
• Given limited tool access, any on‑chain enforcement traces or newer regulatory actions are Not verifiable as of 2026‑08‑27. ### Data protection & ToS risk
• The Spark.fi Terms cover typical website usage, dispute resolution, and disclaimers of liability, but do not read as a regulated financial‑services client agreement.
• Users retain control of assets via a non‑custodial design, but legal recourse is largely limited to the site operator, not the protocol, and governed by offshore law. From an institutional risk perspective, SparkLend should be treated as unlicensed DeFi infrastructure with offshore front‑end operators, geofenced access, and no protocol‑level KYC/AML, implying heightened regulatory and compliance risk for certain jurisdictions.

Yes. For SparkLend on Ethereum, the stablecoin used is USDS, and Spark’s own documentation says the protocol carries depeg risk if USDS loses its $1 peg. However, the results provided do not show a separate, protocol-specific history of USDS depegging, so the exact count for USDS itself is Not verifiable as of 2026-08-26. If you mean the broader stablecoin family relevant to Spark, the provided sources document major depegs for USDC in March 2023 and USDT in May 2022, but those were not SparkLend’s core unit of account. The latest clearly documented major depeg in the provided material is FDUSD on 2025-04-02, which fell to $0.87, a 13% depeg from $1.00. In the same source, USDC’s low was $0.87 (a 13% depeg) and USDT’s low was $0.945 (a 5.5% depeg). So, for the question as stated: USDS depeg history is not verifiable from the supplied sources. If you want, I can next help identify whether Spark switched between DAI / USDC / USDS over time and then assess depeg history for the exact asset used in each phase.

Top 5 risks for SparkLend on Ethereum are: 1. Smart contract risk — Bugs or vulnerabilities in the protocol code could be exploited. Spark’s own FAQ lists smart contract risk as a core risk, and ChainSecurity’s audit still reported one medium-severity issue and a risk-accepted oracle manipulation concern. 2. Oracle / pricing risk — If price feeds are wrong, delayed, or manipulated, collateral can be mispriced and positions can be liquidated incorrectly or bad debt can form. Spark’s docs explicitly frame oracle failure as a failure point that isolation and controls are meant to contain. 3. Liquidation risk — If collateral value falls and health factor drops below threshold, positions can be liquidated; Spark documents partial liquidation above HF 0.95 and full coverage at or below 0.95. 4. Governance risk — Sky Governance configures key market parameters such as supply caps, borrow caps, and kill-switch behavior, so governance capture or malicious proposals could materially change risk settings. 5. Systemic / dependency risk — SparkLend is structurally tied to the broader Sky/Maker ecosystem and its risk controls; external stress, depeg events, or failures in integrated modules can cascade into SparkLend exposure. If you want, I can also turn this into an institutional-style risk memo with severity, likelihood, and mitigants.

SparkLend’s top strengths on Ethereum are: (1) direct access to Sky liquidity, which lets it scale borrowing and lending with strong capital backing; (2) conservative risk design, including a narrow collateral set, multi-oracle pricing, supply/borrow caps, and rate limits; (3) predictable stablecoin borrowing, because the USDS borrowing rate is set by Sky governance rather than short-term pool utilization; (4) non-custodial user control, meaning users retain control of their assets; and (5) a focused collateral base of liquid blue-chip assets, which is intended to reduce liquidation and oracle risk. These are presented in Spark’s documentation and official borrow page for SparkLend on Ethereum.