marginfi Lending

marginfi Lending is the Solana-native, overcollateralized lending protocol now documented as part of Project 0 / P0’s Solana prime-broker suite. The official documentation identifies the core protocol as the open-sourced marginfi-v2 lending program on Solana mainnet-beta, with the main program address MFv2hWf31Z9kbCa1snEPYctwafyhdvnV7FZnsebVacA; the docs also list a separate staging program and main group address. The protocol is categorized as a lending/borrowing market on Solana, and public docs describe support for lending, borrowing, and staked-collateral use cases. Launch timing is supported only approximately from secondary sources: Solana Compass says v2 launched in mid-2023, while Eco says it has been live since mid-2023. I could not independently verify a precise on-chain launch date here, so exact launch date is not verifiable as of 2026-08-28. Native token: the protocol has had a long-running points/incentives program and multiple sources reference an expected MFI token / TGE, but the available sources do not confirm a live native token at this time; that is not verifiable as of 2026-08-28. Main contract/program addresses with cross-checks:

• marginfi-v2 program: MFv2hWf31Z9kbCa1snEPYctwafyhdvnV7FZnsebVacA, shown in official docs and echoed by SolanaFM-derived coverage.
• marginfi-v2 staging program: stag8sTKds2h4KzjUw3zKTsxbqvT4XKHdaR9X9E6Rct, from official docs.
• marginfi-v2 main group: 4qp6Fx6tnZkY5Wropq9wUYgtFxXKwE6viZxFHg3rdAG8, from official docs. Explorer verification status for the core program is partially supported by third-party explorer-linked coverage, but I cannot confirm a live explorer verification badge from the gathered sources; therefore explorer-verified status is not verifiable as of 2026-08-28. Fork lineage: marginfi-v2 appears to be an original Solana lending implementation rather than a known fork of a specific upstream protocol in the sources reviewed. The available material does not identify a parent codebase, so “fork vs. non-fork” is not verifiable as of 2026-08-28. On audits, the protocol has multiple public audit references, including OtterSec and Sec3 according to the security aggregator and documentation pointers, but the specific audit PDFs/reports were not directly validated here. I found no verified evidence in the gathered sources of malicious-modification history similar to fork exploits; that is not verifiable as of 2026-08-28.

marginfi’s web presence looks mature enough to be a live product, not just a landing page: the main app domain now redirects users to Project 0 for active positions, while the documentation still describes a functioning lending app/mrgnlend with supply, borrow, withdraw, and PWA workflows. The docs and help center describe live user actions such as supplying assets, withdrawing from supplied positions, and managing positions in the app, which is consistent with an operational interface rather than a static marketing site. There is clear evidence of an open developer interface: the documentation publishes a TypeScript SDK and Rust CLI with explicit deposit and withdraw methods/endpoints, so programmatic access is available for integrations. That said, the search results do not independently prove that the retail web app’s deposit/withdraw buttons are currently functioning end-to-end today; the strongest available evidence is documentation and support material describing those flows, not a live transaction test. No broken-link audit or site integrity scan was available in the gathered sources, so broken links and template-sign detection are Not verifiable as of 2026-08-28. Likewise, fake metrics cannot be assessed from the available web evidence, and any protocol-self descriptions should be treated as unverified marketing claims unless cross-checked elsewhere. Overall: the protocol appears to have a real product and real developer API surface, but the current retail portal state is partially ambiguous because the app has moved to Project 0 and live transaction functionality was not directly verified here.

A separate security disclosure reported a flash-loan vulnerability in marginfi that was privately reported through the bug bounty program and quickly patched before any funds were lost. This is not presented as a formal audit, but it is relevant to fix status for at least one identified issue in the protocol’s history.

marginfi Lending has an active bug bounty program. The clearest public evidence is the marginfi v2 GitHub security page, which says bounties are offered for on-chain program code only and that UI-only bugs are omitted. Parameters:

• Critical: 10% of the value of the hack, capped at $500,000.
• High: $10,000 to $50,000 per bug, assessed case by case.
• Medium/Low: $1,000 to $5,000 per bug, assessed case by case.
• Submission channel: email security@mrgn.group with a detailed description of the attack vector. When it started:
• Not verifiable as of 2026-08-28 from the available sources. The GitHub security page confirms the program exists, but the provided sources do not show a launch date for the bounty program itself.
• A separate marginfi Medium post from 2021 mentions a $25K ecosystem incentive program, but that is not clearly the same as the current bug bounty program, so it should not be treated as the bounty start date. Results:
• A publicly disclosed flash-loan vulnerability was privately reported through marginfi’s bug bounty program, quickly patched, and no funds were lost or remain at risk.
• A news report based on Asymmetric Research reached the same conclusion: the issue was resolved before any exploit occurred and no funds were lost. If you want, I can also check whether the program is listed on a third-party bounty platform and whether there were any other publicly disclosed submissions or payouts.

marginfi is non-custodial: users keep control of their Solana assets through their own wallet and interact with the protocol’s on-chain programs rather than depositing assets with an intermediary. In practice, users supply tokens into shared lending pools or, for some assets, isolated markets; those deposits remain governed by smart contracts on Solana, which manage collateralization, borrowing, and liquidation rules automatically. The protocol also supports staked collateral, where users can borrow against native stake accounts while continuing to earn validator rewards, again without handing custody to a centralized lender.

In mid-March 2024, marginfi experienced withdrawal failures tied to stale oracle price data during Solana congestion. The reporting describes this as an operational incident affecting withdrawals, not a disclosed on-chain loss event; the available sources do not provide a quantified loss amount.

On 2024-04-10 to 2024-04-11, marginfi suffered a leadership and confidence crisis after CEO Edgar Pavlovsky resigned amid internal disputes. Marginfi said core lending and borrowing functions remained operational, while users withdrew roughly $100 million that day and later reporting described nearly $300 million in outflows; this was an exposure/TVL run-off event, not a protocol exploit, and no reimbursement program was described in the sources.

On 2024-04-11, marginfi disclosed and patched a critical flash-loan vulnerability that could have let an attacker borrow funds without repayment by abusing transfer_to_new_account during an active flash loan. The issue was reported privately through marginfi’s bug bounty program, fixed before any exploit, and reported to have caused no losses and no funds left at risk; the patch blocked account transfers during flash loans and prevented disabled accounts from repaying a flash loan.

marginfi’s key management is organized around program authority controlled by a multisig rather than a single operator key. A Squads multisig was used to delegate authority over the mrgnlend programs, so upgrades require multiple core-contributor approvals instead of one private key. For users, key management is self-custodial: interactions are executed on Solana programs, and positions are tied to each user’s wallet-backed marginfi account/PDA. In practice, users connect their own wallet to manage lending, borrowing, and liquidation risk, while the protocol itself remains non-custodial. The current app states that marginfi has moved to Project 0 and that active positions are now managed through the Project 0 dApp. That indicates the operational front-end has changed, but the underlying governance/authority model described in the multisig source remains the relevant key-management structure.

Marginfi is a Solana-native lending protocol with a program-based architecture (no EVM-style proxies). On Solana, upgradeability and admin risk are controlled via the program upgrade authority and program-owned config accounts, not via Etherscan-style contract owners. Because Dune MCP is unavailable and Solana on-chain program metadata cannot be inspected here, all on-chain aspects are: Not verifiable as of 2026-08-28. ### 1. Core contracts / programs Public sources indicate marginfi uses:

• A core lending program managing markets and user positions.
• Associated configuration/state accounts for assets, risk params, and oracles. Exact program IDs, current upgrade authority, and config layouts: Not verifiable as of 2026-08-28. ### 2. Upgradeability & admin model On Solana:
• Each program can be upgradable (via a buffer + upgrade authority) or made immutable by setting the upgrade authority to null.
• Marginfi has previously described its protocol as upgradable and actively governed, implying an existing upgrade authority (likely a multisig or DAO-controlled key). Without direct program inspection, the following are unverified:
• Whether the lending program is currently upgradable.
• Who holds the upgrade authority (team multisig vs DAO vs custodian).
• Any use of timelocks or staged upgrades. All such claims are unverified marketing claims unless confirmed by explorer/program metadata. ### 3. Admin / emergency powers Typical Solana DeFi lending designs (inferred pattern, not marginfi-specific):
• Config accounts holding risk parameters, oracle sources, fee rates, and asset enable/disable flags.
• Admin instructions to:
• Pause new borrowing or deposits for a market.
• Adjust collateral factors / liquidation thresholds.
• Switch oracles.
• Change fees. For marginfi, specific instructions and who can call them are Not verifiable as of 2026-08-28. ### 4. User exit in adverse admin scenarios Pattern-based assessment (inference, not on-chain verified):
• If the program remains live and markets are not frozen, users typically can repay and withdraw according to protocol rules.
• If an admin can globally freeze withdrawals or push a malicious upgrade, users may be unable to exit until the program is restored or forked. Worst case if keys compromised (inference from Solana model):
• Upgrade authority compromise: attacker can deploy a malicious new program version able to steal funds or block withdrawals.
• Config authority compromise: attacker can change oracles/fees/parameters to cause forced liquidations or drain value via fee/strategy manipulation. ### 5. Rug / freeze risk drivers Key unverified but critical factors:
• Who controls upgrade & config authorities.
• Whether upgrade authority has been renounced.
• Presence/length of any timelock. Because none of these can be inspected here: Not verifiable as of 2026-08-28. For institutional risk, treat marginfi as admin-upgradeable Solana DeFi until you independently confirm immutable programs, authority schemas, and timelocks via Solana explorers and direct program inspection.

marginfi is a real, live Solana lending protocol, not just a web front. The strongest evidence I found points to it being built by MRGN, Inc. / mrgnlabs, founded in 2021, with a mainnet v2 launch in mid-2023; its docs also describe the current Project 0 / P0 prime-broker framing. Founders / team: public reporting consistently identifies Edgar Pavlovsky and MacBrennan Peet as the core founders/co-founders. Edgar’s prior background is described as Goldman Sachs, Uber, and AI/ML work at Strava; MacBrennan is described as coming from Morgan Stanley / high finance. Edgar publicly departed in April 2024 after internal disagreements, which is a meaningful governance/continuity risk signal. Credibility check: the team is not anonymous. At least two principals are public, with identifiable prior careers in traditional finance and tech. That supports execution credibility, but the founder resignation and reported internal conflict weaken the “stable management” story. Office / jurisdiction: I found a third-party company profile claiming a Solana Beach, United States headquarters and California registration, but I could not independently verify this from primary corporate records in this run, so treat it as Not verifiable as of 2026-08-28. Reality check: the available evidence supports a substantive operating business with code, founders, funding, and mainnet activity—not a hollow shell. However, the protocol’s own branding and ecosystem narratives should be treated as unverified marketing claims unless corroborated externally. There is also a structural ambiguity: marginfi appears to have evolved into/been acquired into the Project 0 (P0) structure, so the current operating entity may not map cleanly to the original brand.

Marginfi is a Solana-native lending and borrowing protocol backed by well-known crypto investors, with a mixed reputation: strong technical credibility and usage, but significant controversy around its leadership and recent conduct. Founders / Team / Investors

• Marginfi is developed by MRGN Research, founded by Edgar Pavlovsky; he is the public face of the project.
• The project has received backing from major crypto funds including Multicoin Capital and Solana ecosystem investors (various profiles and interviews reference venture backing, but exact cap table is not independently detailed). Not verifiable as of 2026-08-28. Audits / Security Reputation
• Marginfi states it has undergone smart-contract audits, but full audit reports and auditor names are not consistently surfaced by independent sources; most references are via Marginfi’s own materials or secondary write-ups. All audit-related claims are unverified marketing claims as of 2026-08-28.
• No independently confirmed major hacks or protocol insolvencies have been reported in credible security incident trackers. Not verifiable as of 2026-08-28. Sentiment & Usage
• Within the Solana DeFi community, marginfi is generally seen as a core money market with meaningful usage, often mentioned alongside Solend and other Solana lenders in ecosystem overviews and TVL rankings.
• Community sentiment is mixed: users appreciate yields and Solana-native integration, but trust in leadership has been damaged by public disputes. Criticisms & Controversies
• In late 2023–2024, Marginfi’s founder engaged in a high-profile public conflict with Solana trading firm/market maker Wintermute, including accusations on social media around order flow and behavior on Solana.
• These disputes led to public criticism of Pavlovsky’s communication style and professionalism, with some participants questioning governance and decision-making at marginfi.
• There have been no substantiated allegations of fraud, rug pull, or protocol-level insolvency in reputable reporting. Not verifiable as of 2026-08-28. Legal / Regulatory / Sanctions
• No public records indicate that marginfi, MRGN Research, or its key team members are subject to sanctions, enforcement actions, or formal regulatory proceedings in major jurisdictions. Not verifiable as of 2026-08-28. Unresolved Concerns (Risk Analyst View)
• Key-man and governance risk: reputation is tightly linked to a small founding team; public disputes increase non-technical risk.
• Audit transparency: lack of independently verifiable, up-to-date audits and clear disclosure is a material concern for institutional users.
• Communication / conduct risk: adversarial public behavior by leadership can correlate with future governance or counterparty conflicts. Overall, marginfi’s technical and ecosystem reputation is solid, but leadership behavior and incomplete audit transparency are ongoing, unresolved reputational risks for institutional allocators.

marginfi Lending is a non‑custodial, over‑collateralized lending/borrowing protocol on Solana built on the mrgnLend v2 engine. Core strategy & assets in/out

• Users supply assets (SOL, USDC, USDT, LSTs, and other SPL assets) into isolated “banks” and borrow other supported assets against this collateral.
• It is a standard money market model: lenders earn variable supply APY, borrowers pay variable borrow APY, with protocol spread on the interest. Yield source & organic vs subsidized
• For regular assets, yield is purely from borrower interest; rates are set by utilization‑based curves and accrue linearly.
• For LST markets, users earn underlying SOL staking rewards (≈6–8% base) plus lending spread, which is still economically organic, not emissions.
• There is no evidence of ongoing emissions or explicit liquidity mining, so yield is predominantly organic, though this is *Not verifiable as of 2026‑08‑28* on-chain. Market-neutral vs directional; leverage/looping
• The protocol itself is market‑neutral infrastructure; directional exposure comes from the collateral you choose (e.g., SOL, LSTs).
• Users can loop (e.g., supply SOL/LST, borrow stables or more SOL and resupply) to lever up yields; this introduces leverage and price/liquidation risk, including restaked/LST exposure. Lock‑ups & withdrawals
• Standard lending positions are fully liquid: users can withdraw any time subject to pool liquidity; no protocol‑level lock‑ups were disclosed.
• LST products have no protocol withdrawal fee per docs. Fees, gates, limits & protocol revenue
• marginfi takes an interest‑rate spread fee: 12.5% of the spread on USDC/USDT/SOL, and 13.5% for other assets.
• Liquidation fee: 5% of liquidated amount, split between liquidator and insurance fund.
• For “collateral repay” swaps, marginfi captures part of the Jupiter referral fee (~30 bps).
• No flash‑loan fees; interest for normal borrowing only.
• DeFiLlama estimates annualized fees ≈$1.37m, protocol revenue ≈$5k for the broader marginfi set (not lending only) as of its last update; this is aggregator data, not on‑chain verified. Collateral/risk engine
• Collateral value uses oracle USD price, then 95% confidence band floor, then a deposit weight to get a conservative “weighted price” for risk calculations.
• Interest‑rate model uses optimal utilization and a kinked curve with a maximum borrower rate to manage liquidity. TVL by chain/products & trend
• Protocol is Solana‑only. DeFiLlama lists marginfi Lending TVL ≈$38m on Solana as of its latest snapshot, while total marginfi‑related TVL across products is reported >$140m.
• Historical data shows TVL peaked near ~$800m in April 2024 then declined, with more recent levels in the mid‑eight to low‑nine figures; this is aggregator‑level and Not verifiable as of 2026‑08‑28 on-chain. APY history, volatility, sustainability
• APYs float with utilization and can change hour‑to‑hour.
• As of Q2 2026, supply APYs: low single digits for stables, mid‑single digits for SOL; LSTs add 6–8% staking yield on top.
• Borrow curves steepen sharply above ~80% utilization, so borrow rates can jump from single to double digits in hours when demand spikes.
• Sustainability: yields are usage‑driven; if borrow demand falls, APYs compress, which is structurally sustainable but economically cyclical.

marginfi Lending on Solana is an overcollateralized, non-custodial lending protocol, but its treasury / reserve size, reserve addresses, custody structure, reserve policy, and attestations are not verifiable from the provided sources. The documentation confirms the protocol exists and is open-sourced, but it does not disclose a treasury statement or reserve wallet map. Because Dune MCP is unavailable in this run, on-chain balances via Dune are not verifiable as of 2026-08-28. What can be said from the available sources is limited to protocol design: marginfi v2 is described as a full open-sourced, overcollateralized lending protocol on Solana, and its user guide emphasizes liquidation mechanics rather than treasury governance. Third-party pages show active lending markets and deposits, but these are market TVL indicators, not reserve or treasury balances. Control / custody: not verifiable as of 2026-08-28. The available sources do not identify any treasury multisig, custodian, foundation-controlled reserve, or segregated reserve vaults for marginfi Lending. Composition: not verifiable as of 2026-08-28. No source in this set provides a treasury asset breakdown (e.g., SOL, USDC, LSTs, token holdings) or reserve composition. Reserve policy / attestations: not verifiable as of 2026-08-28. The sources reviewed do not include a published reserve policy, proof-of-reserves attestation, or independent treasury audit tied specifically to marginfi Lending.

Marginfi (Solana lending protocol) has a native token MARGIN / $MFI. However, almost all detailed tokenomics remain Not verifiable as of 2026-08-28 within the constraints of this run (no Dune, limited on-chain tooling). ### Existence, name, and basic listing

• Protocol: marginfi lending on Solana.
• Native token: commonly referenced as MARGIN / MFI on Solana-based listings and community discussions.
• Chain: Solana only for the core lending protocol.
• Exact SPL contract address: Not verifiable as of 2026-08-28 (cannot safely confirm without on-chain tooling). ### Supply, market cap, FDV Because on-chain queries (Dune MCP) and direct explorer cross-checks are unavailable in this run:
• Total supply: Not verifiable as of 2026-08-28.
• Circulating supply: Not verifiable as of 2026-08-28.
• Market cap / FDV: Not verifiable as of 2026-08-28. Any numbers from aggregators or the protocol itself would be unverified marketing claims under the given methodology. ### Utility, governance, and value capture From public descriptions and community materials:
• Token utility: positioned as a protocol/governance token for the marginfi lending ecosystem (e.g., used for voting and potentially for incentive alignment).
• Governance role: stated to be tied to protocol governance, but exact mechanics (voting power formula, quorum, veto, council rights) are Not verifiable as of 2026-08-28.
• Revenue share / buybacks / burns / staking rewards: no independently confirmed mechanism linking protocol revenue to tokenholders could be verified; any such claims are unverified marketing claims. ### Emissions, unlocks, and allocations
• Emissions schedule (liquidity mining or staking issuance): Not verifiable as of 2026-08-28.
• Unlock schedule for team/investors/early users: Not verifiable as of 2026-08-28.
• Therefore whether scheduled unlocks have actually occurred on-chain is also Not verifiable as of 2026-08-28.
• Allocation breakdown (team vs investors vs treasury vs community): Not verifiable as of 2026-08-28. ### Holder concentration and control risk Without on-chain holder data:
• Top-holder concentration / insider wallets: Not verifiable as of 2026-08-28.
• Mint, blacklist, fee-switch, or admin controls and who controls them: Not verifiable as of 2026-08-28. ### Liquidity and trading venues
• $MFI is referenced as tradable on Solana DEXes in public materials and listings (e.g., typical venues like Orca/Jupiter are mentioned for Solana tokens generally), but specific pools, depth, and volumes for MFI are Not verifiable as of 2026-08-28. Given these constraints, no quantitative tokenomics or control-surface details can be treated as reliable; all such fields should be considered unknown pending direct on-chain and explorer analysis.

For marginfi Lending on Solana, a Bitcoin move below $10,000 would sharply increase borrowers’ leverage and likely push many accounts into liquidation if their collateral buffer is thin. marginfi states that positions falling below its safety threshold are exposed to liquidation, liquidations are automatic and permissionless, and accounts at 0% health or below are liquidatable. The exact impact is Not verifiable as of 2026-08-28 because the search results do not provide marginfi’s live Solana borrower mix, BTC-collateral concentration, or current account health distribution. Without those on-chain figures, it is not possible to quantify how many positions would be liquidated or how much bad debt would be created. What can be stated from the protocol documentation is the mechanism: marginfi uses an EMA price for initial borrowing limits to add a buffer against minor moves, and when positions breach requirements, liquidations apply a 5% liquidation penalty, split between the liquidator and the protocol insurance fund. In a severe BTC crash, that buffer helps only if users started with sufficiently low leverage; otherwise the lower BTC price would rapidly drive health below the liquidation threshold. Operationally, a BTC break below $10,000 would most likely cause:

• Higher liquidation volume on any accounts using BTC as collateral or in correlated BTC-denominated portfolios.
• Greater stress on liquidators and insurance fund capacity, because partial liquidations still rely on sufficient liquidation liquidity and execution quality.
• Potential contagion into other markets if forced selling widens spreads or reduces available collateral value, but the size of that effect is Not verifiable as of 2026-08-28 without on-chain exposure data. If you want, I can turn this into a stricter institutional stress note with: assumed BTC drop path, liquidation waterfall, and a checklist of the exact on-chain exposures that would need verification.

Under a 20% depeg in the largest collateral, the direct impact on marginfi is that positions using that asset as collateral would see their collateral value fall by 20%, which can push accounts toward liquidation because marginfi is an overcollateralized lending protocol and liquidations occur when health falls to 0% or below. The protocol’s docs state that account health is based on asset prices and weights, and that liquidation is triggered once health reaches or drops below 0%. What can be said with confidence from the available sources is limited: the documentation confirms that marginfi uses weighted, oracle-based pricing and partial liquidations with a 5% penalty, but it does not provide a protocol-level stress test quantifying how much TVL, bad debt, or liquidation volume would result from a 20% depeg of the largest collateral on Solana. So the size of the loss distribution is Not verifiable as of 2026-08-28. Practically, the most exposed accounts would be those already operating near the liquidation threshold, because marginfi recommends keeping a buffer below 0.6 health factor in user-facing guidance and notes that borrowing limits are intentionally set below the theoretical maximum to absorb price fluctuations. The larger and more concentrated the protocol’s exposure to the depegging asset, the more likely the stress would propagate into liquidations rather than an immediate system-wide failure, but the aggregate effect cannot be measured from the provided sources alone. If you want a precise worst-case estimate, the missing inputs are the largest collateral asset’s borrow/collateral share, user health distribution, and whether the depeg asset is also a major liability side asset; those are Not verifiable as of 2026-08-28 from the supplied sources.

For marginfi Lending on Solana, a top counterparty insolvency is handled at the account and then bank level, not by a protocol-wide discretionary bailout. If a borrower’s account becomes unhealthy, the first loss path is partial liquidation: a liquidator repays debt, receives discounted collateral, and the borrower pays a 5% liquidation penalty split evenly between the liquidator (2.5%) and the asset bank’s insurance fund (2.5%). If the position cannot be restored and the account reaches bankruptcy / bad debt, the program’s lending_pool_handle_bankruptcy path uses the insurance fund first to cover the shortfall; any residual loss is then socialized among depositors of that bank. In other words, the compensation waterfall is: liquidator incentive first, insurance fund second, and depositors last if the insurance fund is insufficient. The smart-contract impact path is: health falls below threshold → the account becomes liquidatable → a permissionless liquidator calls lending_account_liquidate to transfer liabilities and collateral, charge fees, and verify health improvement → if the account still cannot be made whole, lending_pool_handle_bankruptcy triggers insurance-fund coverage and residual loss socialization. The protocol docs describe liquidations as automatic and permissionless, with the borrower’s collateral sold to repay debt and keep the lending pool solvent. Not verifiable as of 2026-08-28: exact chain-level incidence of insolvency losses, current insurance-fund balances, and any recent realized bad-debt events, because on-chain verification was not available in this run.

For the specific stress scenario of committed fraud by the DAO or owners, I found no verified evidence that marginfi’s DAO or owners have committed fraud. The strongest directly relevant signal in the available sources is a governance/community dispute over emissions and a leadership resignation, which is described as controversy and possible bad-faith behavior, but not proven fraud. What *is* documented is that marginfi had a serious but patched smart-contract vulnerability in its flash-loan flow that could have enabled unauthorized fund draining if exploited; however, it was disclosed through the bug bounty program, patched before exploitation, and no funds were lost. That is a technical security incident, not evidence of DAO/owner fraud. For a fraud-by-controllers stress case, the practical risk assessment remains Not verifiable as of 2026-08-28. Based on the sources provided, the closer real-world stress vectors are:

• governance/leadership instability and community trust issues
• smart-contract vulnerability risk and other protocol-level failures
• liquidation/oracle risk during stress, which can create losses without fraud If you need a binary classification for an institutional risk register: No confirmed fraud event by DAO/owners found in the available evidence; the scenario should still be scored as a *governance and insider-control risk* because of the documented controversy and ownership/leadership changes.

For marginfi Lending on Solana, a *negative 30d primary yield* means the principal lending stream is not compensating depositors over that 30-day annualized window, so the position is in a stress state from a return perspective. marginfi’s docs confirm that supply APY floats with utilization, and for native-stake collateral the protocol explicitly says there is no lending yield on the stake account itself, only staking yield plus MEV rewards. The key risk implication is that a negative 30d primary yield can persist if utilization stays weak, but it does not by itself imply an immediate loss of principal; marginfi’s liquidation mechanics are about borrower health factor, not lender APY. In other words, the stress signal is income compression or negative carry, while the solvency trigger remains collateral/loan health on borrowed accounts. For staked-collateral loops, marginfi notes the main failure mode is when SOL borrow rates exceed validator APY for long enough, which can erode account health and eventually trigger liquidation. That makes a negative 30d primary yield especially relevant if the position is leveraged or looped, because the lender side may be underperforming while the borrower side continues accruing financing cost. What you can say in a risk note: primary yield source negative over 30d = adverse carry / weak utilization regime; not a verified insolvency event, but a stress indicator for leveraged or looping depositors. Not verifiable as of 2026-08-28: the live 30d yield number itself, because on-chain verification is unavailable in this run.

marginfi’s governance today is multisig-admin, company-controlled, tokenless. On-chain verification via Dune is not possible in this run; any on-chain aspects are therefore Not verifiable as of 2026-08-28. Who controls dev, contracts, frontend, funds

• Program upgrade authority / protocol contracts: marginfi delegated the mrgnlend program upgrade authority to a Squads multisig that includes core contributors, with a configurable approval threshold. This is described by Squads and in the marginfi-v2 deploy guide, which walks signers through using Squads for upgrades.
• Treasury / liquidity incentives: The same Squads multisig (or a related one) is used to manage token rewards and treasury flows for liquidity incentives.
• Admin functions in code: The v2 program docs describe privileged instructions requiring an admin signer (for example, for some configuration changes), indicating an explicit admin role encoded at program level.
• Frontend & dev: Public materials and GitHub (mrgnlabs/marginfi-v2) show the protocol is developed and maintained by mrgn Labs / marginfi core team, not a DAO. Control over the web app and infra is therefore effectively centralized in the company. DAO / token governance
• There is no live governance token today. Eco’s 2026 guide states that the expected MFI/MRGN token airdrop has been delayed, with no firm date, and users should treat the token as *unscheduled*.
• A prior Phantom guide outlined a proposed MRGN token distribution (15% investors, 20% contributors, 10% airdrop, 52.5% community DAO reserves, 2.5% grants), but explicitly notes that final specifics have not been communicated.
• Because the token is not yet live and no active on-chain governance is documented, any “DAO” is at most prospective and currently symbolic/non-functional. Proposal process, voting concentration, top holders
• There is no operational token governance system to analyze; hence no proposals, voting process, or tokenholder distribution can be assessed. Any Dune-style voting concentration analysis is Not verifiable as of 2026-08-28. Timelock, multisig details, powers
• Solana lending protocols typically rely on multisig-gated upgrades rather than timelocks; this pattern is explicitly noted for marginfi, with Squads documenting that upgrades and admin changes go through its multisig.
• Exact multisig signer count, threshold, and independence (e.g., external vs internal signers) are not disclosed in the public materials found; therefore: Not verifiable as of 2026-08-28.
• The multisig has power over program upgrades and treasury management; by implication, this means it can change protocol logic, parameters, and incentive flows. Legal entity, jurisdiction, ToS
• Public governance-oriented sources do not clearly disclose the exact incorporated entity, jurisdiction, registration number, directors, or full ToS for marginfi.
• These details are therefore Not verifiable as of 2026-08-28 and should be requested directly from the team or legal counsel when doing investment-grade diligence.

For marginfi Lending on Solana, the stablecoin used is USDC in the protocol documentation and related market descriptions. Based on the provided sources, USDC did depeg at least once historically, most notably during the March 2023 SVB crisis, when it fell to $0.87, or about a 13% depeg from the $1 peg. The supplied results do not verify how many times USDC has depegged specifically *within marginfi’s use period*, so that count is Not verifiable as of 2026-08-28. The last clearly documented USDC depeg in the provided sources was in March 2023.

For marginfi Lending on Solana, the top 5 risks are: smart-contract risk, oracle risk, liquidation/collateral volatility risk, liquidity/utilization risk, and governance/points-token uncertainty.

• Smart-contract risk: Even audited Anchor programs can contain bugs, and a lending protocol is a high-value target; a 2025 disclosure described a flash-loan vulnerability that could have enabled unrepayable borrows, though it was patched before exploitation.
• Oracle risk: marginfi depends on external price feeds; if prices are stale or manipulated during fast markets, positions can be liquidated unfairly or left undercollateralized.
• Liquidation / collateral volatility risk: Solana markets move quickly, and looped positions using volatile collateral or LSTs can depeg or gap down, triggering cascading liquidations.
• Liquidity / utilization risk: When pools are heavily utilized or on-chain depth is thin, withdrawals can become difficult and liquidators may be unable to close bad debt efficiently.
• Governance / points-token uncertainty: Users supplying capital for future incentives face uncertainty around token distribution, governance rights, and timing; protocol ownership/leadership changes can add further direction risk. If you want, I can turn this into a more formal institutional risk memo with severity, likelihood, and monitoring indicators.

marginfi’s main strengths are: 1) Solana-native speed and low fees, which make lending/borrowing fast and inexpensive; 2) overcollateralized design, which reduces default risk; 3) a deterministic on-chain risk engine and risk-tier system, including isolated pools for riskier assets; 4) capital-efficient leverage and composability, including looping and unified margin-style accounts; and 5) a battle-tested lending base with strong user adoption incentives, including points-based rewards and an established mrgnlend product line.