Frax (frxUSD / sfrxUSD)

Rejected venues wait the longest for re-review; a rejection has to earn another look before the scheduled date.

REJECTED FOR BOTH SYMBOLS, though the two do not fail for the same reason and one is more reopenable than the other. The prior thesis was factually wrong: current Frax materials do not identify Maple, Ondo or Mountain as frxUSD reserves. Frax says the present backing consists of BlackRock BUIDL and Superstate USTB, while its documentation permits a wider set of cash-equivalent custodian assets and delegates reserve composition to Frax Inc. The more important error was treating frxUSD and sfrxUSD as one risk. frxUSD is a fully collateralized stablecoin redeemable against available enshrined-custodian assets. sfrxUSD is a managed yield vault whose Benchmark Yield Strategy may rotate all staked frxUSD among Ethena carry trades, DeFi AMOs and Treasury/IORB strategies. That is not Treasury exposure and can import venues this registry rejects — sfrxUSD is rejected on that design fact and stays rejected absent a hard mandate exclusion of those venues, a standing condition rather than a temporary evidence gap. frxUSD is different: Frax publishes the permitted designs and current headline backing, but this review could not reproduce a dated, line-by-line reserve report with custodian values and encumbrances — no single Frax-published document gives that, and a Chaos Labs review confirms directly that ”no comprehensive unified attestation covering all frxUSD reserves” exists. This registry has no per-symbol verdict field, so a single entry covering both symbols cannot record ”reject sfrxUSD, hold frxUSD to a materially higher bar than most tokenized-RWA approvals get” without either overstating frxUSD or understating sfrxUSD; a `project` id split was considered and rejected because `project` is the exact join key against the live DefiLlama feed; a fabricated `frax-frxusd` slug would silently never match live pool data. Rejecting the whole entry is therefore the honest position: unproven is not approved. frxUSD alone has a real path back once the schema supports a per-symbol verdict, or once a reproducible reserve report exists and is checked against a standalone entry built for it.

Kill criteria: the events that revoke this verdict
• Any reserve asset outside cash, insured deposits, short government securities, or approved tokenized Treasury funds
• Reserve composition or sfrxUSD allocation not reproducible from dated public reporting
• frxUSD backing below 100% or a custodian redemption failure
• Any sfrxUSD allocation to an unapproved carry, lending, liquidity or private-credit strategy
• Proposed-size frxUSD or sfrxUSD exit exceeding the written cost or timing limit

The research file

The two mechanisms

frxUSD is issued 1:1 against permitted cash-equivalent assets held through “enshrined custodians.” A custodian receives a mint cap and a dedicated contract, mints one frxUSD for one dollar of provable reserves and releases reserve assets when frxUSD is burned. Frax currently names BUIDL and USTB as backing; its docs also describe BUIDL, USTB, JTRSY, WTGXX and AUSD as examples of permitted assets. Those examples are an authority set, not proof of the live allocation.

sfrxUSD is an ERC-4626-like receipt redeemable into frxUSD at an increasing exchange rate. Its Benchmark Yield Strategy is explicitly dynamic: all staked frxUSD may be allocated to the highest-yielding governance-approved carry trade, DeFi AMO or Treasury/IORB strategy. Current docs name Ethena and Superstate for carry, Aave, Curve, Convex, Euler, Fraxlend and dTrinity for DeFi, and BlackRock and Frax Inc/FinresPBC for Treasury/IORB. sfrxUSD is therefore a changing managed portfolio, not merely yield passed through from frxUSD reserves.

Who controls it

Frax DAO retains ownership and may amend or revoke delegation, but FIP-432 gave Frax Inc broad authority over frxUSD compliance, custodian onboarding, bank relationships, collateral composition, audits, attestations and fiat redemption. The proposal states that the company need not obtain an affirmative token-holder vote for each action and that a later DAO vote can revoke or modify the mandate. For sfrxUSD, governance proposals authorize AMOs and strategy additions; the published addresses page identifies AMO operator and timelock multisigs. This is accountable administration, but it is discretionary control over both what backs the stablecoin and where staked balances earn yield. Approval must map the actual signers, thresholds, timelocks and caps, not infer them from the DAO label.

Failure and assurance record

Current frxUSD is a later fully collateralized design and should not be analyzed as if every historical FRAX mechanism still applies. The reviewed primary sources did not disclose a frxUSD reserve shortfall, failed redemption or sfrxUSD principal loss. They also do not provide a complete public incident ledger, so that negative finding is limited. Frax lists July 2025 reviews by Zellic covering frxUSD, the custodian and redemption contracts, and by ChainSecurity covering the sfrxUSD update. Its live bounty pays the lower of 10% of funds at risk or $10M, but only for a working, directly profitable theft or permanent-freeze proof; privileged, frontend and theoretical issues are excluded. Audit and bounty coverage reduce contract risk, not custodian, reserve-valuation, strategy or governance risk.

Exit and liquidity

A holder can redeem frxUSD against whatever reserve asset remains in an enshrined custodian contract. Frax explicitly warns there is no guarantee that a particular asset or custodian will be available after other users consume it. The received asset may itself require whitelisting, so a permissionless contract call does not guarantee a retail holder can complete a dollar exit. Secondary stablecoin pools add another route but must be measured at proposed size.

sfrxUSD can be redeemed without vault-level price impact into frxUSD, but that is only the first leg. Final liquidity still depends on frxUSD reserves and markets, and active strategies may need to unwind during a run. Required tests are the frxUSD available-to-redeem amount by custodian, wallet eligibility for each output, sfrxUSD liquid/encumbered allocation, and executable sfrxUSD-to-USDC cost and time under simultaneous withdrawals.

Comparison and decision recommendation

frxUSD sits between USDC and DAI: it offers multiple tokenized-reserve pathways and on-chain redemption, but reserve and compliance management remain delegated and some output assets are institution-gated. Diversifying BUIDL and USTB can reduce single-fund concentration while adding multiple fund, custodian and access dependencies. sfrxUSD is closer to a managed on-chain yield fund than to sUSDS or a Treasury token because its mandate expressly includes basis trades and DeFi.

Update, 2026-08-16: reject both, for different reasons and different reopen bars. frxUSD is rejected on an evidence gap alone — no reproducible reserve and redemption report exists yet — and reopens the moment one does, ideally through a standalone entry once this registry supports a per-symbol verdict. sfrxUSD is rejected on the design fact itself and reopens only through strategy-level underwriting and hard exclusions of the venues this registry already rejects. The absence of Maple fixes the old memo’s error; it does not convert the combined product into an approval, and holding the whole entry under review while sfrxUSD is already decided delivers nothing an advisor can act on.

Open questions and observable reopen tests

Update, 2026-08-16: no single Frax-published reproducible reserve report was found. What exists instead is fragmented: Chaos Labs’ frxUSD token review names individually audited custodians (Superstate USTB, BlackRock BUIDL with PwC audit and BNY Mellon custody, WisdomTree WTGXX with E&Y audit), but states plainly that ”no comprehensive unified attestation covering all frxUSD reserves” exists, and flags one custodied component (Bridge.xyz/Stripe USDB) whose monthly BPM attestations are conducted but ”not yet publicly disclosed.” Chaos Labs also runs a live Proof-of-Reserves feed (oracles.chaoslabs.xyz/por-feeds/frxusd_por) that was not reachable this pass to confirm current figures. A July 2025 LlamaRisk review found reserves then verified via self-reported spreadsheets rather than SOC-attested reports, with no frxUSD/sfrxUSD contract audit at that time; that finding is stale relative to this review and should be re-checked, not assumed current. No frxUSD depeg or redemption failure was found in 2026. None of this clears the bar for approval; it narrows the gap from ”no evidence” to ”fragmented, partially-audited, not unified,” which keeps allocation at zero pending the actual unified report.

Separately, this entry stores one verdict for two products the thesis already treats differently: sfrxUSD is decided (reject) while frxUSD is genuinely open. The schema currently has no per-symbol verdict, so recording sfrxUSD’s decision without also gating frxUSD requires splitting this into two registry entries — flagged for the next structural pass, not resolved here.

Obtain a dated reserve schedule showing each custodian contract, asset, fair value, mint cap, frxUSD outstanding, encumbrance and redemption eligibility; reconcile it to on-chain balances and an independent attestation. Identify Frax Inc directors, contract signers, thresholds, upgrade authorities and notice periods. For sfrxUSD, publish current allocation, strategy caps, look-through collateral, leverage, counterparties, liquidation terms and realized incident history.

Reopen frxUSD only after monthly independent reserve reporting is reproducible, cash-equivalent backing is at least 100%, no private-credit or basis exposure is present, and a proposed-size path to USDC or dollars remains within the written cost and timing limit. Reopen sfrxUSD only if its mandate hard-excludes every unapproved strategy or each strategy receives its own approval, with a public allocation feed and stress-tested exit. A headline “Treasury backed” label is not an observable test.

Sources

The claims above trace to these. Where a number could not be independently verified, the thesis says so.

• Frax — live frxUSD product and reserve description · primary · accessed 2026-08-14
  Supports: current BUIDL and USTB backing, 1:1 redemption claim, current product design
• Frax docs — frxUSD structure and enshrined custodians · primary · accessed 2026-08-14
  Supports: permitted reserve assets, custodian mint and burn, reserve-asset availability, Frax Inc delegation
• Frax docs — sfrxUSD Benchmark Yield Strategy · primary · accessed 2026-08-14
  Supports: ERC4626-like redemption, Ethena carry strategy, DeFi AMOs, Treasury strategies
• Frax governance — FIP-432 collateral-management delegation · primary · accessed 2026-08-14
  Supports: Frax Inc authority, custodian onboarding, DAO override, annual reporting
• Frax docs — frxUSD and operations contract addresses · primary · accessed 2026-08-14
  Supports: Ethereum and Fraxtal contracts, AMO multisigs, timelock addresses
• Frax docs — published security reviews · primary · accessed 2026-08-14
  Supports: July 2025 frxUSD audit, July 2025 sfrxUSD audit, audit scope
• Frax docs — live bug-bounty terms · primary · accessed 2026-08-14
  Supports: $10M bounty cap, 10% formula, scope exclusions, proof-of-concept requirement

Inherited controls

The verdict above grades the protocol layer. Every position also inherits the asset it holds and the chain it settles on. The least safe layer sets the position’s grade, and the position table names which one that is.