Fluid Lending Protocol

Reassessment Summary (September 2026)

Onchain snapshot at Ethereum block 26079998 (September 29, 2026); Arbitrum and Base reads and DeFiLlama data are from the same day. Core Fluid controls are unchanged: every core contract is still Timelock-owned, the Timelock delay is still 1 day, the Liquidity Layer implementation is the same, and GovernorBravo parameters are the same. Every fToken exchange rate is higher than at the July snapshot. Final score 2.57 → 2.61 (still Medium Risk): Collateralization moves 4.0 → 4.25 because of the USDai hard peg below.

Change #1: sUSDai concentration eased but is still dominant. sUSDai is 27.8% of cross-chain lending TVL ($205.4M), back below the 30% trigger. The drop comes from other assets growing; the sUSDai position itself grew. The per-chain share is 59.6% of Arbitrum and 66.1% of Plasma supply, and sUSDai is also the largest asset on Base (24.3%). sUSDai market cap has grown to ~$487M, so Fluid's ~184M sUSDai shares are now ~42% of circulating supply, down from roughly two-thirds. CoinGecko 24h volume was ~$10.9M (~2.2% of market cap), up from ~$681K in July.

Change #2: every sUSDai vault checked prices USDai at a fixed $1, and ~$214M of debt sits behind those oracles. Across Ethereum, Arbitrum, and Base, each sUSDai vault oracle uses a single hop: the sUSDai→USDai exchange rate, with USDai treated as exactly 1 USDC/USDT/GHO and no market price input. On Ethereum, IGP-136 (July 21) moved all eight sUSDai vaults to FluidChainlinkCappedRate over the Chainlink "SUSDAI / USDAI Exchange Rate" feed. Arbitrum uses a FluidERC4626CappedRate and Base a FluidChainlinkCappedRateL2, both quoting USDai per sUSDai. Onchain debt against sUSDai collateral is ~$106.1M on Ethereum, ~$100.9M on Arbitrum, and ~$7.4M on Base (~$214.3M total), at 76–88% vault-level LTV (Arbitrum vault 66 is ~76.1%) against an 88% collateral factor and 90% liquidation threshold. On the ~88% LTV vaults, a USDai depeg of ~2–3% already puts real LTV past the 90% liquidation threshold without triggering liquidation, and a depeg of ~12% leaves those positions insolvent. The USDai-collateral vaults add ~$4.4M of debt at ~93% LTV against a 94% threshold. USDai traded between $0.9989 and $1.0006 over the 75 days to Sep 29. Plasma could not be checked because no Plasma RPC is configured.

Change #3: TVL recovered. Lending TVL is $737.5M (DeFiLlama, Sep 29), up 15.4% from $639.0M on Jul 14. The 30-day range was $726.1M–$767.3M. Overall Fluid TVL is ~$988.0M.

Other changes (verified onchain):

• Guardian multisig: now 6-of-12, down from 7-of-14. On Sep 9 tx 0xa374…0bff removed two signers and set the threshold to 6. On Sep 13 tx 0x6135…edcd replaced one signer with another. 11 of the 12 current signers were signers at the July snapshot. The threshold stays at half the signer set.
• Governance: proposalCount 135 → 140, and all five new proposals executed (132 of 140 executed in total). IGP-137 moved 5,000,000 FLUID (5% of supply, above the 4M quorum) from the treasury to a dedicated wallet, 0xcabe…3fd7, for the AGI3 institutional-custody partnership. At the snapshot this wallet had not delegated its votes. IGP-138 and IGP-140 cut borrow capacity on legacy collateral (osETH, tBTC, LBTC, ezETH, rsETH, weETHs). IGP-136 and IGP-140 used Liquidity Layer revenue to repay the remaining Resolv-related debt and to fund the Foundation grant.
• sUSDai bridge: on Sep 21, the USD.AI 48h timelock upgraded the Ethereum sUSDai OToken to v1.1 (tx 0x6398…0d41). The adapter address is now hard-coded as the only minter, and a PAUSE_ADMIN_ROLE was added that halts bridge mints and burns (no holder yet on Ethereum). The Arbitrum OAdapter added Solana (Aug 14) and Arc (Sep 10) as peers. That makes five inbound routes, all using the same 3-of-3 DVN set and a 10M sUSDai/hour limiter, that can mint canonical Arbitrum sUSDai.
• Liquidity Layer: no LogUpdateAuths, LogUpdateGuardians, LogPauseUser, or LogUnpauseUser events and no implementation changes since block 25529610. The only rate events were two new listings, USDat (Aug 6) and trUSD (Aug 18). USDC/USDT/GHO/ETH curves are unchanged.
• Cross-references: reUSD's standalone score is now 3.45 (Medium Risk), improved from 3.51. PST grew to $52.0M (7.1% of cross-chain TVL), and Fluid holds ~25.5% of Ethereum PST supply. The PST CCIP pool is still the only minter and Solana is still its only remote chain.
• Audits: no new audits published since the May 24 report.

Contract Addresses (Ethereum Mainnet)

All contracts verified on Etherscan. Compiled with Solidity 0.8.21.

fToken Contracts (Lending)

fToken	Address	Underlying	Underlying Address
fUSDC	0x9Fb7b4477576Fe5B32be4C1843aFB1e55F251B33	USDC	0xA0b86991c6218b36c1d19D4a2e9Eb0cE3606eB48
fUSDT	0x5C20B550819128074FD538Edf79791733ccEdd18	USDT	0xdAC17F958D2ee523a2206206994597C13D831ec7
fWETH	0x90551c1795392094FE6D29B758EcCD233cFAa260	WETH	0xC02aaA39b223FE8D0A0e5C4F27eAD9083C756Cc2
fwstETH	0x2411802D8BEA09be0aF8fD8D08314a63e706b29C	wstETH	0x7f39C581F595B53c5cb19bD0b3f8dA6c935E2Ca0
fGHO	0x6A29A46E21C730DcA1d8b23d637c101cec605C5B	GHO	0x40D16FC0246aD3160Ccc09B8D0D3A2cD28aE6C2f
fsUSDS	0x2BBE31d63E6813E3AC858C04dae43FB2a72B0D11	sUSDS	0xa3931d71877C0E7a3148CB7Eb4463524FEc27fbD
fUSDtb	0x15e8c742614b5D8Db4083A41Df1A14F5D2bFB400	USDtb	0xC139190F447e929f090Edeb554D95AbB8b18aC1C

fToken On-Chain State (Ethereum Mainnet)

fToken	Total Assets (Sep 29 2026)	Total Assets (Jul 14 2026)	convertToAssets(1 share) Jul 14 → Sep 29
fUSDC	131.4M USDC	134.4M USDC	1.205908 → 1.218341
fUSDT	133.2M USDT	130.5M USDT	1.196235 → 1.207877
fGHO	26.66M GHO	15.45M GHO	1.117172 → 1.126087
fwstETH	731.9 wstETH	359.8 wstETH	1.038401 → 1.038476
fWETH	867.0 WETH	1,599.5 WETH	1.077947 → 1.082061
fUSDtb	1.92M USDtb	2.22M USDtb	1.025424 → 1.032478
fsUSDS	13.85 sUSDS	5,015 sUSDS	1.002057 → 1.002057 (no sUSDS borrowing)

Onchain at blocks 25529610 and 26079998. Every fToken exchange rate is monotonically increasing (ERC-4626) and has never decreased at any checkpoint — the key safety property confirming fToken holders have not lost principal value on Ethereum.

Over the Jul 14 → Sep 29 window, fGHO (+72.5%) and fwstETH (+103%) grew, fUSDC (−2.2%) and fUSDT (+2.1%) were roughly flat, and fWETH (−45.8%) and fUSDtb (−13.6%) contracted. fsUSDS was almost fully withdrawn after the Jul 11 pause of the sUSDS-borrowing vaults (tx 0x16aa…c138); its rate is flat because no sUSDS is borrowed.

Core Infrastructure

All core contracts are owned/administered by the Timelock (1-day delay, GovernorBravo admin).

Contract	Address	Role / Key Facts
Liquidity Layer (proxy)	0x52Aa…F4e497	Holds all funds. Upgradeable Instadapp Infinite Proxy; admin = Timelock; current impl 0xcc331…266a2 (since Mar 31 2026).
LendingFactory	0x54B9…51D03	Deploys fTokens; owner = Timelock.
Timelock	0x2386…1F4c	1-day (86,400s) delay; admin = GovernorBravo.
GovernorBravo	0x0204…5fA1B	140 proposals, 132 executed (8 terminal non-executed: 3 Canceled, 3 Defeated, 2 Expired). Quorum 4M / threshold 1M FLUID; 1-day voting delay, 2-day voting period.
Avocado Guardian / Team Multisig	0x4F6F…D49e	6-of-12 custom multisig (7-of-14 until Sep 9 2026); can pause Class-0 protocols and cancel timelock txns; cannot move Liquidity Layer funds. Governance also grants it temporary vault/DEX auth during market launches and routes treasury and revenue transfers through it.
FLUID token	0x6f40…03eb	Governance token; 100M total supply.
Rebalancer (FluidReserveContractProxy)	0x2647…ce92	Permissioned rebalance() on fUSDC/fUSDT; owner = Timelock; deposits underlying as rewards, cannot withdraw.
LiquidityResolver / RevenueResolver	0xca13…5C60 / 0x0A84…0F32	Read-only periphery.

The most powerful admin action is the Timelock upgrading the Liquidity Layer implementation (1-day delay); the current impl is treated as in-scope of the 2025 MixBytes/StateMind Liquidity Layer audits absent a separate post-upgrade review.

Audits and Due Diligence Disclosures

No new audits have been published since the Feb 2026 assessment. The audit landscape is unchanged: 8 distinct security audits across 4 audit firms covering all major protocol components, including the Lending Protocol (PeckShield + StateMind full-protocol audits) and the Liquidity Layer (MixBytes + StateMind dual audit in 2025). All audit reports are available on the audits and security page.

The Resolv USR contagion event was not the result of an unaudited Fluid contract bug; it was the result of a leverage-loop borrower position becoming undercollateralized when an external collateral asset (wstUSR) collapsed in price following the upstream Resolv exploit.

Firm	Date	Scope
PeckShield	Nov 2023	Full Protocol (incl. Lending)
StateMind	Oct–Dec 2023	Full Protocol (incl. Lending)
MixBytes	Mar–Jun 2024	Vault Protocol
Cantina	Sep–Oct 2024	DEX Protocol
MixBytes	Oct 2024	DEX Protocol
MixBytes	Sep–Dec 2025	Liquidity Layer
StateMind	Sep–Oct 2025	Liquidity Layer

No formal verification (Certora, Halmos, etc.) has been performed.

Bug Bounty

Active Immunefi program (under the "Instadapp" name) covering the Liquidity Layer, Lending Protocol, and Vault Protocol — up to $500,000 for critical smart-contract bugs.

Historical Track Record

• Production History: Fluid launched on Ethereum mainnet on February 20, 2024. As of September 29, 2026, the protocol has been in production for ~2.61 years (~952 days).
• Total Fluid TVL (DeFiLlama "fluid" — all products, all chains, supply-side only): ~$988.0M as of September 29 2026 (July 14: ~$771.2M; May 23: ~$999.6M; Apr 27: $911.5M; Feb 2026: $1.45B; peak $2.68B on Oct 9, 2025).
• Lending-only TVL (DeFiLlama "fluid-lending" — all chains): ~$737.5M as of September 29 2026 (July 14: $639.0M; July 6 low: $631.9M; May 23: $872.5M; Apr 27: $750.8M; Feb 2026: $1.28B; peak $2.37B on Oct 9, 2025). Per-chain (September 29): Ethereum $505.1M, Arbitrum $153.0M, Plasma $46.6M, Base $28.4M, Polygon $4.4M.
• Recent TVL trend (DeFiLlama, lending-only daily series): After the May 23 peak of $872.5M, TVL fell to $631.9M on July 6 and has recovered since. Over the 30 days to Sep 29 it stayed between $726.1M and $767.3M. The recovery came from Ethereum (+$79.0M since Jul 14), Arbitrum (+$41.8M), and Base (+$9.9M), while Plasma shrank (−$33.3M).

Major TVL Drawdowns (Historical)

Date	Drawdown	Driver
2024-03-20	-13.7%	Early-protocol churn
2024-08-06	-16.1%	Broader crypto market selloff
2025-04-11	-11.3%	Market stress
2026-03-23	-30.3%	Resolv USR exploit contagion
2026-04-19/20	-17.5%	Kelp DAO bridge exploit / rsETH market freeze

The two 2026 events are the largest single-day drawdowns in the protocol's history. Both were driven by external counterparty/collateral-asset events, not by a bug in Fluid's contracts.

Incidents (since Feb 2026)

1. Resolv USR Depeg / Bad Debt Event — March 22, 2026

Upstream cause: Resolv Labs' minting infrastructure was exploited via a compromised AWS KMS-hosted private key (the SERVICE_ROLE signer). The attacker minted ~80M unbacked USR for ~$200K of USDC collateral, extracted ~$25M, and the USR stablecoin depegged from $1.00 to as low as $0.0025 on Curve before partially recovering to ~$0.85.

Impact on Fluid:

• Concentration risk materialized: wstUSR was the largest single supply asset on Fluid (18.9% of all-chain lending TVL per the Feb 2026 assessment), and ~98% of wstUSR supply was reportedly deployed in Fluid leverage loops at up to 6x leverage.
• Bad debt: Estimated $10–17.5M of bad debt accrued in Fluid lending markets when wstUSR-collateralized borrower positions became insolvent (the wrapping ratio of wstUSR meant that positions were deeply underwater even if USR repegged).
• Outflows: ~$300M of net outflows in a single day (the largest in protocol history). Total Fluid TVL: $1.25B (Mar 22) → $873M (Mar 23) = -30.3% in 24 hours.
• Operational response (~30 minutes): Admin team paused new borrowings and froze new deposits on affected vaults, then began analyzing existing positions. Per Fluid's official statement, all other markets continued operating normally.
• Bad debt coverage: Per Fluid's official communication, 100% of bad debt was covered via a short-term loan coverage agreement funded by personal commitments from Lom Lomashuk (Cyber Fund), a contributor known as "weremeow," and the Fluid core team itself. Resolv Labs additionally executed USR redemptions for whitelisted (pre-incident) wallets covering >90% of the affected user group, and committed to a permanent burn of 46M USR (9M immediate + 36M frozen via blacklist) to reduce ongoing depeg pressure.
• Repayment status: Fluid reported that ~$70M of USR-related debt was fully repaid by March 25, 2026, three days after the incident. All lending markets remained operational. On July 21, 2026, IGP-136 collected Liquidity Layer revenue (USDC, USDT, ETH, GHO, weETH) into the Reserve and forwarded it to the Team Multisig to clear the remaining Resolv-related debt, per the proposal text. The coverage loans were thus partly repaid from protocol revenue.
• fToken impact: No direct loss to fToken holders on Ethereum — onchain exchange rates continued to increase monotonically through and after the event (verified). fToken holders are reported as "not affected" per the official statement.

Sources:

• Halborn — Explained: The Resolv Hack (March 2026)
• Sentora Research — The Resolv Hack: $25M From a Single Compromised Key
• WEEX News — Fluid: 100% of bad debts are covered by the short-term loan coverage agreement
• Phemex News — Fluid Begins $70M Repayments After Resolv Incident
• Protos — Resolv hack shows DeFi learned nothing from last contagion

Critical observation: The bad-debt coverage was discretionary and off-balance-sheet (loans from named individuals/entities). It is not a pre-funded, programmatic insurance fund or first-loss tranche. While the response was rapid and successful, the same coverage mechanism cannot be assumed to scale to a much larger event, and there is no documented contractual obligation forcing those parties to backstop losses again. This is the most material change to the risk profile since the previous assessment.

2. Kelp DAO rsETH Bridge Exploit — April 18, 2026

Upstream cause: Attacker drained 116,500 rsETH (~18% of circulating supply, ~$292M) from Kelp DAO's LayerZero-powered bridge by tricking the cross-chain messaging layer into accepting a forged instruction. This is the largest DeFi exploit of 2026 to date.

Impact on Fluid:

• Fluid froze its rsETH markets within hours, alongside Aave, SparkLend, and Upshift.
• This was a precautionary action — Fluid contracts were not exploited.
• Total Fluid TVL: $1.04B (Apr 18) → $861M (Apr 20) = -17.5% over 2 days.
• No bad debt event reported.

Sources:

• CoinDesk — 2026's biggest crypto exploit: Kelp DAO hit for $292M
• Aave Governance — rsETH Incident Report (April 20, 2026)

Multi-chain Lending Deployment

Per-chain supply is from DeFiLlama (September 29 2026). Utilization is recomputed onchain on Sep 29 from FluidLiquidityResolver getOverallTokenData for every listed token, priced with DeFiLlama coin prices. It covers all Liquidity Layer supply and borrow, including DEX and vault positions. The resulting borrow totals match DeFiLlama's borrowed figures within 0.3% on Ethereum and exactly on Arbitrum and Base. Plasma is carried forward from May 6 because no Plasma RPC is configured.

Chain	Supply	USD-Weighted Util	Highest Borrowed-Token Util (borrow >$0.2M)
Ethereum	$505.1M	~53.9%	GHO 91.2%, ETH 85.4%, USDC 84.7%, USDT 79.8%
Arbitrum	$153.0M	~46.1%	GHO 87.1%, USDC 81.4%, USDT0 79.8%
Plasma	$46.6M	~44.3% (May 6)	USDT0 90.9%, USDe 58.5% (May 6)
Base	$28.4M	~42.2%	USDC 77.0%, GHO 75.6%
Polygon	$4.4M	~28.7%	USDT0 88.5%, USDC 83.9%

Combined lending TVL is $737.5M (September 29), −15.5% from the $872.5M May 23 peak and +16.7% from the $631.9M July 6 low.

Funds Management

How fTokens Work

fTokens are ERC4626-compliant vault tokens. When a user deposits an underlying asset (e.g., USDC), the fToken contract:

1. Calls LIQUIDITY.operate() to deposit the underlying into the Liquidity Layer
2. The Liquidity Layer triggers a callback; the fToken transfers the underlying via SafeTransfer or Permit2
3. Shares are minted to the user based on the current exchange rate

On withdrawal, the reverse occurs: shares are burned before the underlying is withdrawn from the Liquidity Layer (burn-first pattern for safety).

Exchange rate: Computed onchain as tokenExchangePrice / EXCHANGE_PRICES_PRECISION (1e12 precision). The rate is monotonically increasing — it can never decrease. Verified onchain on Apr 27, 2026: all fToken exchange rates have increased since Feb 2026.

It incorporates:

• Yield from the Liquidity Layer (borrower interest)
• Optional rewards from a LendingRewardsRateModel (currently inactive for all fTokens; yields are purely organic)

Safety mechanisms in fToken contracts:

• Custom reentrancy guard (deposit/withdraw/rebalance all protected)
• Callback validation: checks caller = Liquidity AND token = ASSET AND status = ENTERED
• Burn-before-withdraw pattern
• BigMath precision with SafeCast overflow protection
• Rewards rate capped at 50% APR maximum

Accessibility

• Supplying: Permissionless — anyone can deposit via fTokens. No whitelist required.
• Redemption: fToken withdrawals via withdraw() or redeem() (standard ERC4626). Subject to Liquidity Layer withdrawal limits. During the Mar 2026 event, withdrawals from affected (wstUSR-related) vaults were paused while solvent markets continued operating; standard fToken markets (fUSDC etc.) on Ethereum remained operational throughout.
• Fees: No explicit deposit/withdrawal fees. Interest rates are algorithmically determined by utilization via a kink-based model.

Yield Source and Counterparty Risk

fToken yield comes from borrower interest. Borrowers use the Vault Protocol to deposit collateral and borrow assets from the Liquidity Layer. This means fToken holders are exposed to:

• Vault Protocol solvency: If borrowers default and liquidations fail to recover full value, bad debt could affect lending reserves. The Mar 2026 event is a concrete example of this risk materializing — the wstUSR price collapse outpaced the liquidation engine's ability to safely close positions, producing $10–17.5M of bad debt.
• Liquidation effectiveness: The tick-based liquidation mechanism must function correctly to prevent bad debt accumulation. In the Mar 2026 case, the speed and magnitude of the wstUSR price collapse exceeded what the liquidation mechanism could handle without losses.
• Oracle correctness: Vault liquidations depend on Chainlink, UniswapV3 TWAP, and Redstone price feeds. Oracle failures could delay liquidations.
• Coverage mechanism: Bad debt is not covered by a programmatic, pre-funded insurance fund or first-loss tranche. The Mar 2026 incident was covered via discretionary short-term loans from named individuals and entities.

Collateral quality backing fToken yield (borrower collateral types):

• Blue-chip: ETH, WETH, wstETH, weETH, rsETH, WBTC, cbBTC
• Stablecoins: USDC, USDT, USDT0, sUSDe, GHO
• Yield-bearing / higher-risk (the concentration drivers): sUSDai / USDai (27.8% — USD.AI synthetic dollar, Arbitrum-native and bridged), PST (7.1% — Huma PayFi RWA, bridged from Solana via CCIP), reUSD (5.4% — Re Protocol, Medium Risk 3.45), plus smaller positions such as sUSDe, and previously wstUSR (substantially de-risked post-Mar 2026). The sUSDS borrow markets were paused on Jul 11 2026.

Collateralization

• Backing: All lending positions are over-collateralized onchain. Borrowers must maintain collateral ratios (80–95% LTV depending on the pair).
• Liquidations: Fully onchain tick-based mechanism. Liquidation penalty as low as 0.1% for correlated pairs (wstETH/ETH), higher for uncorrelated pairs.
• Withdrawal Gap: Extra gap on Liquidity Layer limits reserved for liquidations to ensure they can always execute.
• Limitation observed Mar 2026: the liquidation engine's effectiveness depends on the collateral asset behaving like a liquid market asset. When wstUSR's underlying USR depegged ~99.7% intraday, on-DEX liquidity for wstUSR was insufficient for safe liquidation, producing bad debt.

Provability

• Transparency: All reserves are fully onchain and verifiable via resolver contracts (FluidLiquidityResolver at 0xD7588F6c99605Ab274C211a0AFeC60947668A8Cb).
• Exchange Rate: fToken exchange rates are computed programmatically onchain (ERC4626 standard). No offchain oracle or admin input needed. Rate is monotonically increasing — verified.
• Interest Rates: Algorithmically determined based on utilization. USDC rate model (re-verified onchain May 6, 2026): kink at 85% utilization (5.40% rate), second kink at 93% (7.50%), max rate 40%. Proposal #128 has now executed, but it did not emit a USDC/USDT rate update and the live USDC/USDT curve remains unchanged.
• Revenue: Protocol revenue is calculated and verifiable via the RevenueResolver contract.

Interest Rate Model

Decoded from FluidLiquidityResolver.getTokenRateData(token) for each token ((uint256 version, RateDataV1Params v1, RateDataV2Params v2)). Compared against the May 6 and Feb 2026 snapshots in prior reports. All curves unchanged since May 24.

Stablecoins and ETH — current curves:

Token	Version	Kink 1	Rate@K1	Kink 2	Rate@K2	Max Rate
USDC	V2	85%	5.40%	93%	7.50%	40.00%
USDT	V2	85%	5.40%	93%	7.50%	40.00%
GHO	V2	85%	6.50%	93%	9.50%	40.00%
ETH (native)	V2	88%	2.50%	93%	4.00%	10.00%

Rate-update events on the Liquidity Layer since May 24, 2026: the only LogUpdateRateDataV2s events were new-token listings: PST on May 18, USDat on Aug 6 (tx 0x3ab0…4e7d), and trUSD on Aug 18 (tx 0x5c10…8305). Both new listings use a 50%/80% kink curve with a 100% max rate. USDC/USDT/GHO/ETH curves unchanged since May 6.

Post-Feb USDC/USDT rate history (verified that no new events since the April 23 2026 update — rechecked at block 26079998):

Date	Tx	USDC/USDT Kink 1	Rate at Kink 1	Kink 2	Rate at Kink 2	Notes
Feb 13, 2026	0xe373...131bb	85%	5.00%	93%	8.00%	Liquidity Layer rate update event
Mar 10, 2026	0xa99e...96c06	85%	4.50%	93%	7.50%	Liquidity Layer rate update event
Apr 23, 2026	0x1927...49e	85%	5.40%	93%	7.50%	Liquidity Layer rate update event; still the current live USDC/USDT curve

Proposal #128 (executed May 5 2026) was expected by its text to move USDC/USDT kinks to 90%/95%; the execution receipt did not emit a USDC/USDT rate event and the live state did not change. This unresolved gap between proposal text and onchain effect remains a documented open observation; no follow-up proposal has corrected it.

Liquidity Risk

Lending-Specific Liquidity Concerns

fToken holders face liquidity risk from the shared Liquidity Layer architecture. This architecture and its trade-offs are unchanged since Feb 2026, but the Mar 2026 event provides a concrete stress test.

• Shared pool: fToken withdrawals compete with all other withdrawal demand on the Liquidity Layer.
• Withdrawal limits: The Liquidity Layer enforces per-token expandable withdrawal limits. maxWithdraw() returns the minimum of: (1) the withdrawal limit at Liquidity, (2) actual liquid balance.
• Stress test result (Mar 22–25, 2026): $300M+ net outflows in 24 hours. Standard fToken markets (fUSDC, fUSDT, etc.) on Ethereum continued processing withdrawals throughout. Affected vaults (wstUSR-collateralized) were paused. The kink-based rate model worked as designed — high utilization produced rate increases that incentivized borrowers to repay and stabilize utilization.
• Stress test result (Apr 18–20, 2026): Additional ~$180M outflows over 2 days following the Kelp/rsETH freeze. No further bad debt; precautionary freeze of rsETH markets only.

Exit Mechanisms

• Normal exit: Call withdraw() or redeem() on fToken. Subject to available liquidity and withdrawal limits.
• Secondary market: fTokens are ERC20 tokens and can be traded on secondary markets, though no significant DEX liquidity for fTokens was observed.
• Throttled exit: During high utilization, the expansion-rate mechanism throttles large withdrawals.

Lending TVL by Asset Type (Ethereum)

Source: DeFiLlama fluid-lending Ethereum tokensInUsd (Sep 29, 2026): $505.1M of supply routed through the Liquidity Layer. This includes Vault-collateral deposits, not only fToken supply. Percentages are of that Ethereum total.

Asset Type	Supply TVL	%
ETH/LSTs (wstETH, WETH, osETH, weETH, rsETH, …)	~$202.6M	40.1%
Yield-bearing stablecoin wrappers (sUSDai, reUSD, sUSDe)	~$116.3M	23.0%
Stablecoins (USDT, USDC, GHO, trUSD, USDe, …)	~$69.1M	13.7%
BTC tokens (WBTC, cbBTC, eBTC, LBTC)	~$56.2M	11.1%
Other (PST, FLUID, PAXG, XAUT, …)	~$61.0M	12.1%

Yield-bearing stablecoin wrappers are ~23% of Ethereum supply, still the second-largest category after ETH/LSTs. sUSDai on Ethereum is $76.5M (15.1% of Eth supply), up from $64.6M on Jul 14. Fluid's Ethereum Liquidity Layer holds 68.55M of the 85.13M Ethereum sUSDai OToken supply (~80.5%, onchain at block 26079998). Ethereum sUSDai is a bridged representation whose source chain is Arbitrum, not a native Ethereum ERC-4626 vault; the burn/mint authority on both chains is documented below. PST alone is $52.0M (10.3% of Ethereum supply).

Top Supply Assets (Cross-Chain)

Source: DeFiLlama fluid-lending tokensInUsd — cross-chain, total $737.5M (Sep 29, 2026).

Rank	Token	Supply TVL	% of Total
1	sUSDai	$205.4M	27.8%
2	wstETH	$158.3M	21.5%
3	PST	$52.0M	7.1%
4	USDC	$47.2M	6.4%
5	WBTC	$42.5M	5.8%
6	reUSD	$39.7M	5.4%
7	USDT	$35.7M	4.8%
8	WETH	$31.7M	4.3%
9	USDT0	$26.2M	3.6%
10	cbBTC	$24.6M	3.3%
11	osETH	$16.4M	2.2%
12	weETH	$14.4M	2.0%
13	ETH	$8.5M	1.2%

Top-5 concentration: 68.6% (69.0% on Jul 14, 70.6% on May 23). Top single-asset concentration: 27.8%, below the 30% trigger (31.2% on Jul 14, 30.6% on Jul 6, 28.3% on May 23, 19.9% on Apr 27). sUSDai + reUSD: 33.2% of cross-chain TVL in yield-bearing stablecoin wrappers (37.3% on Jul 14). Including PST, assets whose value depends on offchain credit or admin-reported NAV make up 40.3%.

Top Supply Assets (Ethereum)

Source: DeFiLlama fluid-lending Ethereum tokensInUsd — total $505.1M (Sep 29, 2026).

Rank	Token	Supply TVL	% of Total
1	wstETH	$135.4M	26.8%
2	sUSDai	$76.5M	15.1%
3	PST	$52.0M	10.3%
4	reUSD	$38.7M	7.7%
5	USDT	$35.7M	7.1%
6	WBTC	$34.2M	6.8%
7	WETH	$30.9M	6.1%
8	USDC	$26.7M	5.3%
9	cbBTC	$19.0M	3.8%
10	Other	$56.0M	11.1%

Top-5 concentration: 67.0%. Top single-asset concentration: 26.8% (wstETH). Higher-risk assets sUSDai + PST + reUSD combined: $167.2M (33.1%) of Ethereum supply.

Non-blue-chip collateral beyond the wrappers — PST (Ethereum rank 3, $52.0M, 10.3%). Unlike Fluid's blue-chip collateral (wstETH, WETH, WBTC, cbBTC), PST is Huma Finance's "PayFi Strategy Token". It carries three stacked risks that blue-chip assets do not (mint path first verified onchain Jul 22 and re-verified Sep 29, 2026). Fluid's Ethereum Liquidity Layer holds 45.80M of the 179.30M Ethereum PST supply (~25.5%). IGP-136 raised the PST/USDC and PST/USDT T1 vault max borrow limits from $10M to $15.1M.

• Off-chain RWA credit backing: PST represents a claim on Huma's PayFi receivables (real-world payment-financing / invoice credit), not onchain collateral — its value is not independently verifiable onchain and depends on the performance of off-chain receivables.
• Chainlink CCIP bridge: on Ethereum PST is a BurnMintERC20 (0x22ae3d9a…d4c7) — mint authority is the CCIP BurnMintTokenPool (0xBE77…0D3d), the only MINTER_ROLE holder, with no role grants since May 9 2026. A CCIP compromise could mint unbacked PST on Ethereum. The pool's inbound rate limiter from Solana has a 7.7M PST capacity and refills at ~6,112 PST/s, so a full bucket refills in about 21 minutes.
• Solana source-chain risk: the CCIP pool's only configured remote chain is Solana (getSupportedChains() returns solely the Solana selector 124615329519749607), so PST is effectively bridged from Solana and inherits Solana-side program/custody risk.

PST was listed May 18 2026 with a steep 50%/80% kink and 100% max rate — Fluid's own rate curve treats it as a higher-risk market than the stablecoin/ETH pairs.

Per-Chain Concentration (Worse Than Cross-Chain Average) — September 29, 2026

Chain	Total Supply	#1 Asset	#1 Share	#2 Asset	#2 Share
Plasma	$46.6M	sUSDai	66.1%	USDT0	30.3%
Arbitrum	$153.0M	sUSDai	59.6%	USDC	10.9%
Base	$28.4M	sUSDai	24.3%	wstETH	21.7%
Polygon	$4.4M	WBTC	33.5%	wstETH	30.4%
Ethereum	$505.1M	wstETH	26.8%	sUSDai	15.1%

On Arbitrum and Plasma, sUSDai (issued by USD.AI, a synthetic-dollar protocol backed by AI hardware loans) still dominates lending TVL (59.6% / 66.1%, down from 63.9% / 75.3% on Jul 14). A USD.AI-level upstream event would still functionally take down lending on those two chains. sUSDai is now also the largest asset on Base ($6.9M, 24.3%), where Fluid holds 6.18M of the 6.33M Base sUSDai supply (onchain Sep 29). On Ethereum it is $76.5M (15.1%).

Issuer risk profile (USD.AI): USD.AI (usd.ai) describes itself as "a yield-bearing synthetic dollar backed by loans against AI hardware, compute, and DePIN assets" targeting 15–25% APR, with peg maintenance relying on arbitrage rather than over-collateralization or RWA backing. Per CoinGecko on Sep 29, sUSDai has a ~$487M market cap (~436.6M circulating), trades at ~$1.12, and had ~$10.9M of 24h volume (~2.2% of market cap). The July snapshot showed ~$681K (~0.2%); a single-day volume figure is noisy. Fluid's Liquidity Layers hold ~184M sUSDai shares: 68.55M on Ethereum, 81.70M on Arbitrum, and 6.18M on Base (onchain), plus ~27.6M on Plasma (derived from DeFiLlama's $30.8M). That is ~42% of circulating supply, down from roughly two-thirds in July. Fluid is still the largest single holder and venue, so a large exit would still face material price impact. On-chain price history: ATL $0.796 on Sep 26 2025 (−20% from par), confirming sUSDai is not a stable-value instrument.

sUSDai cross-chain architecture and LayerZero mint authority (re-verified Sep 29, 2026): The canonical sUSDai contract is the Arbitrum StakedUSDai (implementation version 1.13, 0xef07…860d; the Arbitrum proxy was upgraded on Jul 28, Jul 31, and Sep 21 2026 through the same ProxyAdmin), an ERC-4626 vault whose asset() is Arbitrum USDai 0x0A1a…82EF. Its bridge mint/burn functions are gated to an immutable bridge-adapter address. BRIDGE_ADMIN_ROLE was revoked on Apr 29 2026 when the check moved into the implementation. The same address on Ethereum, 0x0b2b…955ef9, is an upgradeable OToken with no asset() / convertToAssets() path. The 48h USD.AI timelock upgraded it on Sep 21 2026 (tx 0x6398…0d41) to implementation 0x192a…01fe (OToken v1.1). In v1.1, mint/burn are restricted to the immutable OAdapter, and a PAUSE_ADMIN_ROLE can halt both. No account had been granted that role on Ethereum at the snapshot; on Arbitrum it is held by a 2-of-3 Safe, 0x3a32…b872.

The same-address OAdapter 0xffB2…7f24 burns on send and mints on receive on both chains, so its authenticated receive path can mint canonical Arbitrum sUSDai. The Arbitrum adapter has five inbound peers, and each can trigger that mint:

Source	Peer set	DVNs required	Confirmations	Hourly limit
Ethereum	May 27 2025	3-of-3 (LayerZero Labs, Nethermind, Canary)	15	10M
Plasma	Sep 19 2025	same 3-of-3	20	10M
Base	Nov 18 2025	same 3-of-3	10	10M
Solana	Aug 14 2026 (tx 0x116d…902b)	same 3-of-3	32	10M
Arc	Sep 10 2026 (tx 0xcd1e…f98bc)	same 3-of-3	10	10M

The Ethereum-side receive route from Arbitrum is a separate 3-of-3 DVN set with 20 confirmations. Both adapters are owned by the same-address 3-of-3 Safe 0x5F0B…841F, which has identical owners on both chains; see the Bridges page. Upgrades on both chains go through ProxyAdmin 0x0b32…8d9F, owned by the 48-hour TimelockController. The 10M sUSDai/hour limiters apply to outbound debit and do not cap _credit on a forged inbound message. A verifier-path or adapter-owner compromise could dilute canonical supply and impair Fluid's cross-chain exposure independently of Fluid's own vault oracles. Each new peer adds another source chain whose compromise could reach that path. The uniform 3-of-3 DVN quorum and 3-of-3 owner Safe are meaningful mitigants, but they do not remove the canonical-supply bridge dependency.

sUSDai vault oracles (Ethereum, Arbitrum, Base; verified Sep 29, 2026): IGP-136 (Jul 21) moved all eight Ethereum sUSDai vaults (171–173, 175–179) and the sUSDai-USDC / sUSDai-USDT DEX center prices to oracles built on FluidChainlinkCappedRate 0xC5D2…76a0. That contract reads the Chainlink SUSDAI / USDAI Exchange Rate feed, which returned 1.11525 at the snapshot. The T1 oracles, e.g. GenericOracle_SUSDAI_USDC 0x08E9…Ef4e, have one hop: that capped rate divided by 1e12 for decimals. USDai is therefore priced at exactly 1 USDC/USDT/GHO, with no market price input. The other chains use the same structure. The Arbitrum sUSDai oracles (e.g. 0x603E…5042) have one hop, FluidERC4626CappedRate 0x1428…6c2d ("USDAI per 1 SUSDAI"). The Base oracle 0xB1f5…1bc4 has one hop, FluidChainlinkCappedRateL2 0x6318…5496.

Debt and LTV per vault, read from VaultResolver.getVaultEntireData on each chain on Sep 29 (all 182 Ethereum, 95 Arbitrum, and 50 Base vaults scanned for sUSDai/USDai collateral). LTV is vault-level debt divided by collateral at the oracle price. Smart-debt (T3/T4) debt is DEX borrow shares valued at the oracle-implied share price (~$2.257 on Ethereum, ~$2.152 on Arbitrum). All vaults use CF 88% / LT 90%. Dust vaults are omitted.

Chain	Vault	Type	Debt	LTV
Ethereum	sUSDai / USDC (171)	T1	$21.90M	84.7%
Ethereum	sUSDai / USDT (172)	T1	$21.93M	87.4%
Ethereum	sUSDai / USDC-USDT smart debt (173)	T3	$15.21M	86.8%
Ethereum	sUSDai-USDC / USDC-USDT (175)	T4	$4.47M	87.9%
Ethereum	sUSDai-USDT / USDC-USDT (176)	T4	$2.12M	87.9%
Ethereum	sUSDai-USDT / USDT (177)	T2	$20.41M	85.4%
Ethereum	sUSDai-USDC / USDC (178)	T2	$18.66M	87.8%
Ethereum	sUSDai / GHO (179)	T1	$1.34M	85.9%
Arbitrum	sUSDai / USDC (65)	T1	$33.49M	82.5%
Arbitrum	sUSDai / USDT0 (66)	T1	$25.73M	76.1%
Arbitrum	sUSDai-USDC / USDC (67)	T2	$15.61M	87.9%
Arbitrum	sUSDai / USDC-USDT0 smart debt (68)	T3	$10.52M	83.9%
Arbitrum	sUSDai-USDT0 / USDT0 (69)	T2	$15.49M	87.5%
Base	sUSDai / USDC (47)	T1	$4.83M	87.1%
Base	sUSDai-USDC / USDC (48)	T2	$2.56M	86.8%

Total: ~$106.1M on Ethereum, ~$100.9M on Arbitrum, ~$7.4M on Base, ~$214.3M overall of stablecoin debt secured by sUSDai at a fixed USDai = $1. On the maxed T2/T4 vaults (~88% LTV), a USDai depeg of only ~2–3% already puts real LTV above the 90% liquidation threshold without triggering liquidation. A depeg past ~12% leaves those vaults' debt larger than their collateral value; the threshold is ~18% for the lowest-LTV large vault, Arbitrum 65. USDai traded between $0.9989 and $1.0006 over the 75 days to Sep 29 (CoinGecko); its all-time low was $0.770 (Sep 26 2025). The Plasma sUSDai vaults were not checked because no Plasma RPC is configured (TODO).

External corroboration: per CoinDesk RWA Yield Infrastructure Trade (Mar 2026), Fluid handled ~100% of on-chain sUSDai trading volume and 68% of reUSD trading volume at that time, so a redemption-side stress event in sUSDai would also concentrate on Fluid's liquidity venues.

Concentration Risk Reassessment (September 29, 2026)

sUSDai crossed the 30% trigger (set May 24) on Jul 6 and peaked at 31.2% on Jul 14. At this snapshot it is 27.8% of cross-chain lending TVL. Its absolute size is slightly higher ($205.4M vs $199.2M); the share fell because wstETH, PST, USDC, and WBTC supply grew faster.

Key facts:

• sUSDai cross-chain share is still 8.9 pp higher than the wstUSR share that preceded the Mar 2026 incident.
• On Arbitrum and Plasma, sUSDai remains a single point of failure for the chain's lending business (59.6% / 66.1% of chain TVL).
• Two of the top 6 supply assets (sUSDai #1, reUSD #6) are yield-bearing stablecoin wrappers, the same structural amplification pattern as wstUSR/USR. Together they are 33.2% of all-chain TVL. PST, an offchain-credit RWA, adds another 7.1%.
• Fluid's share of total sUSDai supply fell to ~42%, and reported 24h sUSDai volume rose to ~2.2% of market cap. Fluid is still the largest single sUSDai holder.
• USDai is priced at a fixed 1:1 in every sUSDai vault oracle checked, so a USDai depeg would not reach liquidation pricing on ~$214.3M of debt across Ethereum, Arbitrum, and Base (~29% of lending TVL).

The structural reasoning holds: yield-bearing stablecoin wrappers carry contagion risk because their wrap ratio amplifies losses when the underlying depegs. Share-based concentration improved, but because USDai is fixed at $1 in the sUSDai oracles, a USDai-level event would not be caught by liquidations on any chain checked.

Historical Liquidity Performance

• August 2024: -16.1% TVL drop, recovered without operational issues
• March 2026: -30.3% drop, partial market freezes (wstUSR vaults), bad debt event covered, full recovery within days. Standard fToken markets remained functional.
• April 2026: -17.5% drop, precautionary rsETH freeze, no operational impact to other markets.
• May–Jul 2026: TVL fell 27.6% from peak to low ($872.5M → $631.9M on Jul 6). The decline was broad-based across all chains and asset types, with no incident and no bad debt. Targeted LogPauseUser actions paused a USDai/USDC vault on Jun 14 (tx 0xbde0…4ebb) and three sUSDS-borrowing vaults on Jul 11 (tx 0x16aa…c138) as market wind-downs. fToken exchange rates kept increasing throughout.
• Jul–Sep 2026: TVL recovered to $737.5M (Sep 29) with no incidents, no pauses, and no bad debt.

Centralization & Control Risks

Governance

• Governance Model: Onchain GovernorBravo governance. FLUID token holders vote on proposals that execute through a timelock. Discussion on governance forum, onchain voting via GovernorBravo, and offchain signaling via Snapshot.
• Timelock: 0x2386DC45AdDed673317eF068992F19421B481F4c — 1-day (86,400s) delay. Admin = GovernorBravo.
• Owner/Admin: All core contracts (Liquidity Layer proxy admin, LendingFactory, RebalancerProxy) confirmed owned by the Timelock.
• GovernorBravo Parameters:
  • Quorum: 4,000,000 FLUID (4% of total supply) ✓
  • Proposal threshold: 1,000,000 FLUID (1% of total supply) ✓
  • Voting delay: 7,200 blocks (~1 day) ✓
  • Voting period: 14,400 blocks (~2 days) ✓
  • Proposals created: 140; proposals executed: 132 (other 8 are terminal non-executed: 3 Canceled, 3 Defeated, 2 Expired; verified by iterating state() at block 26079998)
• Token distribution change (IGP-137): 5,000,000 FLUID moved from the Treasury DSA to the Team Multisig on Aug 7 2026. On Aug 14 it was forwarded (tx 0x2d08…d9ef) to a dedicated contract wallet, 0xcabe…3fd7, for distribution to institutional custodians under the AGI3 partnership. The proposal says these tokens are legally locked until 2030, but that lock is not enforced onchain. At the snapshot the wallet held all 5M FLUID and had not delegated its votes. 5M FLUID exceeds both the 4M quorum and the 1M proposal threshold, so delegation of this block would be a material shift in voting power. The proposal text also says Kinetic Group plans to acquire up to 10% of FLUID supply on the open market.

Lending-Specific Admin Controls

Role	Who	What They Can Do to Lending
Timelock (governance)	0x2386DC45...	Upgrade Liquidity Layer implementation, change LendingFactory owner, change supply/borrow configs, change rate models
LendingFactory Auths	Set by Timelock	Update fToken rewards config, change rebalancer address, rescue stuck tokens, set fToken creation code
LendingFactory Deployers	Set by Timelock	Create new fToken contracts
Rebalancer (fUSDC, fUSDT)	0x264786EF…ce92 — FluidReserveContractProxy, owner()=Timelock	Deposit underlying without minting shares (adds as rewards). Cannot withdraw. (Corrected from Apr 2026 report, which named the inactive FluidLendingRewardsRateModel 0x724d…b9b6 in this slot.)
Guardian (Avocado multisig, 6-of-12)	0x4F6F977a...	Pause Class 0 protocols only. Cannot move or withdraw Liquidity Layer funds. Cancel timelock transactions. Used to freeze rsETH markets in Apr 2026 (precautionary). Governance temporarily grants it vault/DEX auth during market launches (e.g. IGP-138, IGP-140) and removes that auth after launch.

Key finding: No admin role can directly access or move user funds deposited via fTokens. The most powerful action is the Timelock upgrading the Liquidity Layer implementation (1-day delay). The Guardian's pause capability was exercised appropriately in both Mar and Apr 2026 events without abuse. The last Liquidity Layer pause was on Jul 11 2026 (sUSDS-borrowing vault wind-down). No LogPauseUser, LogUnpauseUser, LogUpdateAuths, or LogUpdateGuardians events occurred between block 25529610 (Jul 14) and block 26079998 (Sep 29). The guardian multisig changed from 7-of-14 to 6-of-12 on Sep 9 2026 and had a one-for-one signer replacement on Sep 13. The threshold stays at half the signer set.

Programmability

• System Operations: Largely programmatic. Interest rates and exchange rates are all computed onchain algorithmically.
• Oracle System (dependency via Vault Protocol): Chainlink primary, with UniswapV3 TWAP, Redstone, and custom center-price oracles as fallbacks. Modular per vault.
• Rate Model: Interest rates determined algorithmically via kink-based utilization model. Parameters set by governance.
• Keepers/Automation: No keepers needed for lending. Liquidations (in Vaults) are incentivized and performed by external liquidators.

External Dependencies

• Liquidity Layer: Critical dependency — holds all fToken deposits. Upgradeable proxy controlled by Timelock.
• Vault Protocol: Generates fToken yield. Vault borrowers, liquidations, and oracles all affect lending counterparty risk.
• Chainlink: Indirect dependency via Vault Protocol oracle system. Multiple fallback oracle paths reduce risk.
• Permit2: Supported for deposits (Uniswap's 0x000000000022D473030F116dDEE9F6B43aC78BA3).
• External collateral asset issuers: As demonstrated by the Resolv (USR/wstUSR) and Kelp (rsETH) events, the protocol's risk surface includes the operational security of every accepted collateral asset issuer. A compromise of an issuer's keys or bridge can produce contagion damage even without any bug in Fluid.

Operational Risk

• Team: Instadapp Labs. Founded by Sowmay Jain and Samyak Jain — both are publicly known, India-based founders active since 2019. Key GitHub contributors include thrilok209, KABBOUCHI, and SamarendraGouda.
• Funding: Well-funded by top-tier VCs: Pantera Capital, Coinbase Ventures, Standard Crypto, additional undisclosed investors.
• Legal Structure: Instadapp Labs.
• Documentation: Comprehensive technical documentation at docs.fluid.instadapp.io. Full source code on GitHub.
• Communication: Active governance forum, Discord, Twitter @0xfluid.
• Incident Response: Mar 2026 response demonstrated the team can act within ~30 minutes (pause/freeze affected markets) and arrange off-balance-sheet capital coverage of ~$70M within ~3 days. Strong response, but reliance on personal commitments rather than a programmatic mechanism is a structural concern.

Monitoring

Contracts to Monitor

Contract	Address	What to Monitor
fUSDC	0x9Fb7b4477576Fe5B32be4C1843aFB1e55F251B33	Second largest fToken (~$131M Sep 29 2026). Exchange rate, deposits/withdrawals
fUSDT	0x5C20B550819128074FD538Edf79791733ccEdd18	Largest fToken (~$133M Sep 29 2026). Exchange rate, deposits/withdrawals
Liquidity Layer	0x52Aa899454998Be5b000Ad077a46Bbe360F4e497	Holds all fToken deposits. Admin changes, implementation upgrades
Liquidity Layer impl (current)	0xcc331daf69752bece3dc98dbc63eacd5092266a2	Implementation contract behind the proxy. Monitor for changes via EIP-1967 implementation slot.
Timelock	0x2386DC45AdDed673317eF068992F19421B481F4c	Owner of all core contracts — queued/executed transactions
GovernorBravo	0x0204Cd037B2ec03605CFdFe482D8e257C765fA1B	Governance proposals, voting, execution
Avocado Multisig	0x4F6F977aCDD1177DCD81aB83074855EcB9C2D49e	Guardian pause/cancel actions; requiredSigners() (6) and signer-set changes
AGI3 FLUID allocation wallet	0xcabebc7f76d53582a4be7d9972a2b4f531753fd7	5M FLUID; DelegateChanged / transfers (would exceed quorum if delegated)
sUSDai capped-rate oracle	0xC5D27C5d356479b681328351F1583c63051E76a0	Source for all Ethereum sUSDai vault oracles; compare against the USDai market price, because the USDai leg is fixed at 1:1
Arbitrum sUSDai rate source	0x14288700dc560a3f26a81ff10ab19c8b75846c2d	FluidERC4626CappedRate for all Arbitrum sUSDai vault oracles; same USDai = $1 assumption
sUSDai-collateral vaults (all chains)	See the sUSDai vault oracle table in the Liquidity Risk section	Aggregate debt (~$214.3M at Sep 29: Ethereum ~$106.1M, Arbitrum ~$100.9M, Base ~$7.4M) and vault-level LTV vs the 90% liquidation threshold

Key Events to Watch

Contract	Event	Significance
Timelock	QueueTransaction / ExecuteTransaction	Governance actions queued/executed — 1 day warning
LendingFactory	New fToken creation	New lending market created
Liquidity Layer	LogUpdateAuth	Auth permissions changed — affects who can modify lending configs
Liquidity Layer	LogUpdateGuardian	Guardian address changed
Liquidity Layer	LogPauseUser / LogUnpauseUser	Protocol paused/unpaused — directly affects fToken operations. Now an actively exercised path (Mar/Apr 2026).
Liquidity Layer	LogUpdateUserSupplyConfigs	Supply limits changed — affects max fToken deposits
Liquidity Layer	LogUpdateUserBorrowConfigs	Borrow limits changed — affects utilization and withdrawal availability
Liquidity Layer	LogUpdateRateDataV1 / LogUpdateRateDataV2	Interest rate parameters changed — affects fToken yield
EIP-1967 Admin (proxy)	Storage slot read	Implementation changes on Liquidity Layer / fToken contracts

New Monitoring Recommendation: Collateral Asset Issuers

Given the Mar/Apr 2026 contagion events, monitoring of the off-protocol collateral asset issuers (Resolv, Kelp, Ethena, Maple, etc.) is now a first-order concern. A compromise at any major upstream issuer can produce sub-day bad-debt events at Fluid even without any Fluid contract change.

Top-priority issuer to monitor (September 2026 update): USD.AI (sUSDai / USDai). sUSDai is 27.8% of cross-chain Fluid lending TVL ($205.4M; below the 30% trigger but still the largest asset) and 59.6% / 66.1% of Arbitrum/Plasma supply. sUSDai is issued by USD.AI (usd.ai). USD.AI is a synthetic-dollar protocol backed by AI hardware loans with peg maintenance via arbitrage, targeting 15–25% APR. Key risk factors: Fluid holds ~42% of circulating sUSDai supply. Price has historically ranged $0.796–$1.19. USDai peg stability is now directly monitorable risk, because every sUSDai vault oracle checked prices USDai at a fixed 1:1, with ~$214.3M of debt behind them. Also monitor the canonical sUSDai mint path: new OAdapter peers, DVN or confirmation changes, PAUSE_ADMIN_ROLE grants, and proxy upgrades on either chain. Any incident touching USD.AI governance, the underlying loan-collateral pool, or USDai redemption mechanics is the single highest-impact external event for Fluid lenders at this snapshot.

Second-priority: Huma / PST (bridged from Solana). See below; PST is now the #3 cross-chain asset.

Third-priority: Re Protocol (reUSD). reUSD is $39.7M cross-chain (5.4% of cross-chain TVL). On Ethereum, Fluid's Liquidity Layer holds 35.07M of the 250.80M reUSD supply (~14.0%, onchain Sep 29). The standalone risk assessment at reports/report/re-reusd.md scores 3.45/5.0 (Medium Risk) as of Sep 23, 2026. Key risk factors for Fluid lenders: roughly one-third of NAV is backed offchain (onchain reserve coverage ~66.7% reUSD-only). The share price is written daily by a Chainlink Functions NAVConsumer; its admin now sits behind a 48h Timelock. ~94% of onchain reserves are held at three plain EOAs, and ~99.5% of onchain reserves are sUSDe. There is no bug bounty and branch test coverage is 42%. Any incident touching Re's custodial EOAs, the NAVConsumer oracle, the §114 trust counterparties, or Re's governance Safe and Timelock would affect the reUSD position on Fluid.

Lower priority: Sky/Maker (sUSDS, fsUSDS). The sUSDS-borrowing vaults were paused on Jul 11 2026, and fsUSDS supply is ~14 sUSDS.

Huma / PST (bridged from Solana) — detail. PST (Huma "PayFi Strategy Token") is $52.0M (7.1% of cross-chain TVL, all on Ethereum) and is not blue-chip: it is backed by off-chain PayFi receivables rather than onchain collateral, and on Ethereum is a Chainlink CCIP BurnMintERC20 (0x22ae3d9a…d4c7) bridged solely from Solana (pool 0xBE77…0D3d supports only the Solana chain selector). Monitor the CCIP token-pool mint activity and rate limits, Huma's receivables performance and redemption liquidity, and any Solana-side program incident — any of which could impair or unbacked-mint the Ethereum PST that Fluid vaults hold as collateral.

Reassessment Triggers

• Time-based: Reassess in 2 months (November 2026). This is shorter than the standard 6 months because of the ~$214M sUSDai hard-peg exposure and the growing sUSDai bridge surface.
• Concentration-based: The 30% sUSDai trigger fired on Jul 6 2026 (30.6%, peak 31.2% on Jul 14); sUSDai is now 27.8%. Reassess immediately if sUSDai exceeds 40% of cross-chain lending TVL, or if sUSDai + reUSD + PST exceed 50%.
• Coverage-mechanism formalization: Reassess if Fluid implements a programmatic insurance/coverage layer (would lower funds-management subcategory). Conversely, reassess if the protocol experiences a second bad-debt event without comparable third-party coverage.
• TVL-based: Reassess if lending TVL changes by more than 50% from the ~$737.5M baseline (September 29 2026), i.e., drops below ~$369M or exceeds ~$1.106B.
• sUSDai source-chain / bridge controls (highest priority): Reassess immediately on any of the following on Arbitrum or Ethereum sUSDai: a mint-role or admin-role change, a PAUSE_ADMIN_ROLE grant, an OAdapter owner change, a new peer or DVN-route configuration change, a rate-limit change, a proxy implementation upgrade, a cross-chain supply mismatch, or an unexpected canonical Arbitrum mint.
• USDai depeg / hard-peg debt: Every sUSDai vault checked (Ethereum 171–173 and 175–179; Arbitrum 65–69; Base 47–48) and every USDai-collateral vault prices USDai at a fixed 1:1. Monitor the USDai market price and the debt in these vaults. Reassess immediately if USDai trades below ~$0.98, if total hard-peg debt exceeds ~$300M (currently ~$218.7M including USDai vaults), or if their oracle configuration changes. A ~2–3% USDai depeg pushes the most leveraged sUSDai vaults (~88% LTV) past their 90% LT without liquidation. Past ~12% they become insolvent.
• Incident-based: Reassess after any further exploit, governance change, significant parameter modification, or contagion event from a major collateral issuer (especially USD.AI for sUSDai/USDai, Huma for PST, Re for reUSD).
• Governance: Reassess if GovernorBravo parameters change (quorum, timelock delay, voting period), if the Avocado guardian threshold or signer set changes, or if the IGP-137 5M FLUID allocation wallet delegates its votes or is distributed to a party that can reach quorum alone.
• Dependency: Reassess if Liquidity Layer implementation is upgraded (current impl: 0xcc331daf69752bece3dc98dbc63eacd5092266a2) or if a new protocol is added to the shared liquidity pool.

Open Observations

• Proposal #128 (executed May 5 2026) didn't change USDC/USDT kinks despite the proposal text saying it would; only an ETH rateAtUtilizationMax event was emitted. All subsequent proposals (#129–#140) have since executed. No USDC/USDT rate event has been emitted since Apr 23 2026 (rechecked at block 26079998), so the curves remain at 85%/93% kink with 5.40%/7.50% rates.
• TODO — Plasma: the Plasma sUSDai vault oracles and their debt, and Plasma utilization, were not verified because no Plasma RPC is configured. Plasma holds $30.8M of sUSDai supply. Source: VaultResolver.getVaultEntireData and the oracle getOracleHopSources() at 0xA5C3…95cC on Plasma.

Appendix: Contract Architecture

┌─────────────────────────────────────────────────────────────────────────┐
│                          GOVERNANCE LAYER                                │
│                                                                          │
│   FLUID Token Holders   (100M max supply, 4M quorum, 1M propose)         │
│              │                                                           │
│              │ propose / vote                                            │
│              ▼                                                           │
│   ┌────────────────────────────┐      ┌──────────────────────────────┐  │
│   │  GovernorBravo             │      │  Avocado Multisig (Guardian) │  │
│   │  0x0204Cd03...             │      │  0x4F6F977a...               │  │
│   │  proposalCount: 140        │      │  Custom contract (not Safe)  │  │
│   │  132 Executed              │      │  6-of-12 custom guardian     │  │
│   │  voting: 1d delay, 2d vote │      │  - Pause Class-0 protocols   │  │
│   │                            │      │  - Cancel timelock txns      │  │
│   └────────────┬───────────────┘      └────────────┬─────────────────┘  │
│                │ queue                             │ cancel              │
│                ▼                                   ▼                     │
│   ┌──────────────────────────────────────────────────────────────────┐  │
│   │  Timelock — 1-day delay (admin = GovernorBravo)                  │  │
│   │  0x2386DC45AdDed673317eF068992F19421B481F4c                      │  │
│   │  Owns: Liquidity Layer admin slot, LendingFactory, VaultFactory, │  │
│   │        DexFactory                                                │  │
│   └──────────────────────────────────────────────────────────────────┘  │
└──────────────────────────────────┬──────────────────────────────────────┘
                                   │ owns / upgrades
                                   ▼
┌─────────────────────────────────────────────────────────────────────────┐
│                        CORE INFRASTRUCTURE                               │
│                                                                          │
│   ┌──────────────────────────────────────────────────────────────────┐  │
│   │  Liquidity Layer  (Instadapp Infinite Proxy)                     │  │
│   │  Proxy: 0x52Aa899454998Be5b000Ad077a46Bbe360F4e497                │  │
│   │  EIP-1967 dummy impl (current): 0xcc331daf… (since Mar 31 2026)  │  │
│   │      previous: 0xa57d7cEF…  (upgrade tx 0xf484b2a2…)             │  │
│   │  Module-dispatch logic in module slots, NOT EIP-1967 impl slot   │  │
│   │                                                                  │  │
│   │  Holds ALL deposits across lending / vaults / DEX / stETH        │  │
│   └────┬─────────────────────────────────────────────────────────────┘  │
│        │                                                                 │
│   ┌────▼────────────────┐  ┌─────────────────┐  ┌───────────────────┐   │
│   │  LendingFactory     │  │  Resolvers      │  │  Sibling Factories│   │
│   │  0x54B91A0D...      │  │  FluidLiquidity │  │  VaultFactory     │   │
│   │  Deploys fTokens    │  │  Revenue        │  │  DexFactory       │   │
│   └────┬────────────────┘  └─────────────────┘  └────────┬──────────┘   │
└────────┼───────────────────────────────────────────────────┼────────────┘
         │ deploys                                           │ deploys
         ▼                                                   ▼
┌──────────────────────────────────┐    ┌────────────────────────────────┐
│   LENDING PROTOCOL (fTokens)     │    │   SIBLING PROTOCOLS            │
│   ERC4626, monotonic exch. rate  │    │   (share Liquidity Layer)      │
│                                  │    │                                │
│   fUSDC   0x9Fb7…  ~$131M        │    │  Vault Protocol                │
│   fUSDT   0x5C20…  ~$133M        │    │   - borrowers + collateral     │
│   fGHO    0x6A29…  ~$27M         │◀───│   - tick-based liquidations    │
│   fwstETH 0x2411…  ~732 wstETH   │ yield   - generates fToken yield    │
│   fWETH   0x9055…  ~867 WETH     │    │                                │
│   fUSDtb  0x15e8…  ~$1.9M        │    │  DEX Protocol                  │
│   fsUSDS  0x2BBE…  ~14 sUSDS     │    │  stETH Protocol                │
│                                  │    │                                │
│   All-chain lending TVL: ~$738M  │    │                                │
└──────┬──────────────────▲────────┘    └─────────┬──────────────────────┘
       │ deposit            ▲ withdraw            │ borrow / repay
       ▼                    │                     ▼
   Lenders               (no value loss        Borrowers
                          Mar/Apr 2026)        (post collateral)
                                                  │
                                                  ▼
┌─────────────────────────────────────────────────────────────────────────┐
│                       EXTERNAL DEPENDENCIES                              │
│                                                                          │
│  ORACLES  (Vault Protocol consumes these for liquidations)               │
│    Chainlink (primary)  •  UniswapV3 TWAP  •  Redstone  •  custom        │
│                                                                          │
│  COLLATERAL-ASSET ISSUERS  ← first-order contagion vector                │
│    Top-10 cross-chain supply (Sep 29 2026, total $737.5M):               │
│      sUSDai   27.8%  (USD.AI — Staked USDai)            ★ (peak 31.2%)   │
│      wstETH   21.5%  (Lido)                                              │
│      PST       7.1%  (Huma PayFi RWA, CCIP from Solana)                  │
│      reUSD     5.4%  (Re Protocol — score 3.45 Medium)  ★                │
│      USDC      6.4%   WBTC  5.8%   USDT  4.8%   WETH  4.3%               │
│      USDT0     3.6%   cbBTC 3.3%                                         │
│    ★ = yield-bearing stablecoin wrapper (same structural pattern as      │
│        wstUSR/USR that produced the Mar 22 2026 bad-debt event)          │
│    sUSDai: issuer USD.AI; Fluid holds ~42% of supply. Per-chain:         │
│        Plasma 66.1%, Arbitrum 59.6%, Base 24.3%, Ethereum 15.1%.         │
│    All sUSDai vault oracles fix USDai = $1 (~$214M debt, 3 chains).      │
│                                                                          │
│    Materialized contagion (since Feb 2026 assessment):                   │
│      Mar 22 2026 — Resolv USR depeg → wstUSR collateral collapse →       │
│                    $10–17.5M bad debt (covered off-balance-sheet)        │
│      Apr 18 2026 — Kelp DAO bridge exploit → rsETH precautionary freeze  │
│                    (no Fluid contract loss)                              │
│                                                                          │
│  PERMIT2 (Uniswap):  0x000000000022D473030F116dDEE9F6B43aC78BA3          │
└─────────────────────────────────────────────────────────────────────────┘

Risk pathways (key):

• Lender → fToken → Liquidity Layer: fully programmatic ERC4626 with burn-before-withdraw; exchange rate monotonically increasing (verified through all 2026 events including the May–Jul TVL drawdown).
• Borrower default → bad debt → fToken yield: realized in Mar 2026 ($10–17.5M). No programmatic backstop; covered via discretionary loans (cyberfund/Lomashuk, weremeow, Fluid core team).
• Collateral-issuer / bridge compromise → wrap-ratio amplified loss → bad debt: the structural pattern that produced Mar 2026. sUSDai is 27.8% cross-chain and 59.6% / 66.1% on Arbitrum/Plasma, still larger than the wstUSR exposure that preceded the prior incident. Issuer and source chain: USD.AI on Arbitrum, a novel synthetic-dollar protocol backed by AI hardware loans. Fluid holds ~42% of circulating sUSDai, and sUSDai has traded as low as $0.796. The LayerZero OAdapter has mint/burn authority on the Ethereum OToken and canonical Arbitrum sUSDai. Five source chains (Ethereum, Plasma, Base, Solana, Arc) reach the canonical mint path, each through the same 3-of-3 DVN quorum, and the outbound rate limiters do not cap inbound credit. reUSD (5.4%, $39.7M) has a standalone score of 3.45/5.0 (Medium Risk) per reports/report/re-reusd.md: offchain backing component, Chainlink-Functions-written share price, EOA reserve custody, no bug bounty. sUSDai + reUSD make up ~33% of TVL, and PST another 7.1%.
• USDai depeg → no liquidation on sUSDai vaults → bad debt: every sUSDai vault oracle checked (Ethereum, Arbitrum, Base) fixes USDai at $1, behind ~$214.3M of debt at 76–88% vault-level LTV (90% LT). A ~2–3% USDai depeg pushes the most leveraged vaults past the liquidation threshold on real prices, and past ~12% they become insolvent, with no liquidation trigger in either case.
• Governance → Timelock (1d) → Liquidity Layer impl: upgrade path. The Mar 31 2026 upgrade bundled dummy-impl swap + module-dispatcher selectors in one tx (0xf484b2a2…). No subsequent upgrades (rechecked Sep 29 2026).
• Avocado Guardian (6-of-12) → Pause Class-0: cannot move Liquidity Layer funds; used appropriately in both Mar and Apr 2026 events. The last pause was the Jul 11 2026 sUSDS-vault wind-down, with no pause/unpause events since.