# Adversarial critique B: omissions, selection skew, stale figures and unmodelled stress **Reviewed:** `artifacts/report.md` (SHA-256 `1a7edfe5…bb4d49`, 85 claims C001–C085, gaps G01–G13) and `artifacts/data.json` (SHA-256 `4cf8e25e…c49ccc17`). **Review date:** 2026-10-29. **Scope of this critique:** what is missing or skewed. Citations were not re-checked one by one. I did not rewrite the report. **Labels used here.** **Fact** means a source located during this review states it. Most of these sources are secondary (press, auditor blogs), and they are named where used. **Inference** means my analysis. **Uncertain** means I could not confirm it in the time available and it needs checking before anyone relies on it. Every case put forward to fill a gap is a lead for the authors to verify against primary sources. None has been verified at primary level here. --- ## 0. Verdict in one paragraph The report is careful about the claims it does make. It labels its claims, separates issuer statements from verified figures, and is honest that it has no calibrated debt ceiling. Its weakness is the **case selection**, which is skewed in two directions that matter for imdUSD: 1. It never looks at the **Liquity lineage that imdUSD is copying**. That means BOLD's own live record, Liquity-v2 forks backed by an ecosystem token, and Liquity-v1 forks that failed. 2. It treats Uniswap v4 hooks only through a single launch-stage, issuer-documented success story (StablePair). It omits the two hook-based protocols that lost funds through hook logic in 2025. The own-token section (Q2) is also missing the case closest to "a CDP stablecoin lent against the issuer ecosystem's governance token with thin depth": **crvUSD/LlamaLend and other lending exposure to CRV around Michael Egorov's positions (July 2023 and June 2024)**. The report discusses the June 12, 2024 crvUSD incident only as keeper latency. The design section's ceiling and solvency formulas (C076, C077) treat collateral depth L as independent of the stablecoin's own stress. For a small stablecoin backed by its own token, that is the main failure mode, and the report does not model it. --- ## 1. Omitted protocols and incidents inside the brief's scope ### 1.1 Q1 and Q5: the Liquity-v2 reference design is described from docs, never from its record The report cites Liquity v2 only through documentation and the repository (C070, C071). It does not report: - **The February 2025 Stability Pool issue and redeployment.** *Fact (Liquity blog, [stability-pool-issue](https://www.liquity.org/blog/stability-pool-issue)):* - Liquity was informed of a potential issue on Feb 12, 2025 and confirmed it on Feb 13. - The issue "could result in the loss of funds held in these contracts" (the Stability Pools). - Users were told to close positions, and the first deployment became "legacy". - A five-week audit contest and re-audits followed, and v2 relaunched on **May 19, 2025** ([liquity-v2-redeployment](https://www.liquity.org/blog/liquity-v2-redeployment), [liquity-v2-is-live](https://www.liquity.org/blog/liquity-v2-is-live)). The design imdUSD is adopting therefore had to be wound down once in its first months. The report's "reference mechanism" framing hides this. *Inference:* the migration away from a legacy deployment is itself a peg-stress event: borrowers were told to close loans and holders without a loan were told to sell. That makes it relevant to Q3-style "near-misses" and to imdUSD's own upgrade plan. - **BOLD's measured peg, redemption volume and borrower-rate behaviour since May 2025.** *Uncertain / gap:* - No BOLD price series, redemption totals or rate-distribution data appear anywhere in the report. - C070 claims borrower-set rates "encourage rates to adjust when demand weakens". That is a design claim made without the record of the design. Liquity's own [interest-rate management post](https://www.liquity.org/blog/interest-rate-management-in-liquity-v2) and Dune dashboards for BOLD redemptions are where the evidence is. - **Liquity-v2 forks backed mainly by an ecosystem's own network token.** The brief asks for "any other stablecoin overcollateralised mainly by its own governance or network token since 2023". **Felix feUSD** on HyperEVM is the closest live analogue to imdUSD and is absent. *Uncertain (aggregator and audit sources only):* - It is a permitted Liquity-v2 fork with HYPE, kHYPE and UBTC collateral, borrower-set rates and lowest-rate-first redemptions. - Its admin and pause controls go beyond Liquity's. - A Three Sigma price-feed review found the HYPE/USD feed implicitly assumed USDC = $1 ([Three Sigma case study](https://threesigma.xyz/case-studies/lending/felix-protocol)). The report should have asked: - what share of feUSD backing is HYPE or kHYPE; - how feUSD traded during HYPE drawdowns, including October 10–11, 2025; - how redemptions and Stability Pool absorption performed against a token whose depth sits mostly on one venue. Other Liquity-v2 forks that may list governance tokens (for example Nerite on Arbitrum) should be checked for the same reason. *Uncertain:* I did not confirm their collateral lists. ### 1.2 Q3 and Q5: Liquity-family and CDP failures are missing, so the selection is skewed toward exotic synthetics Section 4 covers Stream, Elixir, USDX, Resolv, Ethena, Usual and Resupply. It omits CDP failures in imdUSD's own architecture class: | Case | What happened (label) | Why it matters for imdUSD | Lead source | |---|---|---|---| | **Raft R**, Nov 10, 2023 | *Fact (press):* attacker borrowed 6,000 cbETH, exploited a mint bug in Raft's wrapped-collateral contract, minted ~6.7M R and sold it through Balancer/Uniswap for ~$3.3–3.6M. R fell to ~$0.18. Protocol sunset; recovery plan ~42%. Audited (Trail of Bits, Hats). | A Liquity-style CDP whose supply invariant failed at the **collateral-wrapper** layer, not in the core trove logic. Directly relevant to imdUSD wrapping **staked** IMD. | [The Block](https://www.theblock.co/post/262681/defi-platform-raft-pauses-minting-of-its-stablecoin-after-acknowledging-security-vulnerability), [Rekt](https://rekt.news/raft-rekt/), [Blockworks](https://blockworks.co/news/exploit-ether-defi-protocol) | | **Prisma mkUSD/ULTRA**, Mar 28, 2024 | *Fact (post-mortem via search):* MigrateTroveZap flaw; ~3,257 ETH (~$11M) lost; protocol paused. | Liquity-fork failure in **migration/zap periphery**. imdUSD is "adding more collateral types", which means migrations and new branch periphery. | [Prisma post-mortem](https://hackmd.io/@PrismaRisk/PostMortem0328), [CertiK](https://www.certik.com/resources/blog/prisma-finance-incident-analysis) | | **Abracadabra MIM**, Jan 30, 2024 | *Fact (press):* ~$6.5M cauldron exploit; MIM to ~$0.76–0.77; DAO treasury buyback-and-burn announced; partial recovery to ~$0.94–0.98. Repeated exploits in 2025 should be checked (*uncertain*). | A CDP stablecoin with no PSM that defended its peg with a **treasury buyback**. That is the "own-ecosystem resources as last-resort bid" pattern the report never tests. | [Cointelegraph](https://cointelegraph.com/news/abracadabra-s-6-49m-loss-leads-to-mim-stablecoin-destabilization), [Rekt](https://rekt.news/abra-rekt/) | | **Yala YU**, Sept 14, 2025 | *Fact (press):* 120M unauthorized YU minted; ~$7.6–7.7M extracted; YU to ~$0.2046; again ~$0.42 in November 2025. | Second unauthorized-mint case beside Resolv. It shows Resolv (C053–C056) is a **pattern**, and that the peg did not recover months later. | [The Block](https://www.theblock.co/post/370547/yalas-bitcoin-backed-stablecoin-yu-depegs-after-security-breach-leads-to-unauthorized-mint), [Cointelegraph](https://cointelegraph.com/news/yala-yu-stablecoin-depegs-after-attack) | | **Falcon USDf**, Jul 8, 2025 | *Fact (press), lows conflict:* reported lows $0.8871 (DEX Screener via The Defiant) to $0.978 (CMC). ~$25M of reserves on-chain vs ~$609M off-chain. 7-day, whitelisted redemption. Altcoin collateral including thin tokens. Back near par by Jul 10. | A synthetic dollar that **survived**, with thin-token collateral and gated redemption. The report has no survivor of this type except Ethena. | [Cointelegraph](https://cointelegraph.com/news/falcon-usd-stablecoin-depegs-liquidity-collateral-concerns), [The Defiant](https://thedefiant.io/news/tokens/dwf-labs-usdf-stablecoin-briefly-depegs-amid-doubts-over-collateral-and-yield), [DL News](https://www.dlnews.com/articles/defi/dwf-ceo-pledges-asset-breakdown-as-stablecoin-breaks-peg/) | *Uncertain, out of scope for a reserve-free design but relevant to C079's supply-bound recommendation:* Paxos's accidental very large PYUSD mint in October 2025, reversed within minutes. The authors should verify the details before citing it. ### 1.3 Q2: the strongest own-ecosystem-token-collateral case is CRV, and it is missing - **July 30 – August 2023, Vyper/Curve exploit.** *Fact (press, figures conflict):* - Egorov's CRV-backed loans across Aave, Fraxlend, Inverse and Abracadabra were reported at ~$100–168M. - They were backed by roughly a third of circulating CRV. - CRV fell ~$0.73 → ~$0.50 between Jul 30 and Aug 1. - The Aave position was reported at ~63.2M USDT against ~305M CRV, with a liquidation price ~$0.377. - Positions were reduced through OTC sales (one report: 39.25M CRV for 15.8M USDT), because on-chain CRV depth could not absorb a liquidation. Sources: [CoinDesk 2023-07-31](https://www.coindesk.com/markets/2023/07/31/a-curve-founders-168m-stash-is-under-stress-creating-a-risk-for-defi-as-a-whole), [Cointelegraph](https://cointelegraph.com/news/curve-liquidation-risk-poses-systemic-threat-to-defi-even-as-founder-scurries-to-repay-loans), [Blockworks](https://blockworks.co/news/aave-curve-exposure). - **June 12–13, 2024.** *Fact (press, figures conflict):* - About $95.7M of stablecoin debt was reported against ~$141M CRV across five protocols. - About $50M of it was crvUSD on LlamaLend, where three Egorov accounts were >90% of borrowed crvUSD. - CRV fell ~30%. - Reported bad debt was ~$10–11.5M, repaid mostly by Egorov plus a third party. Sources: [The Block](https://www.theblock.co/news/defi/2024-06-13-curve-founder-loan-positions-liquidation-risk-299864), [CoinDesk](https://coindesk-coindesk-prod.cdn.arcpublishing.com/markets/2024/06/13/crv-slides-30-as-loans-tied-to-curves-founder-face-liquidation-risk), [The Defiant](https://thedefiant.io/news/defi/curve-finance-founder-michael-egorov-suffers-massive-liquidations). - **Why this is a skew, not just an omission (Inference).** C009 attributes the June 12, 2024 crvUSD **upward** depeg to PegKeeper-v2 gate latency and spillover liquidations. It does not connect the event to the CRV liquidations happening at the same time. *Uncertain:* the precise causal link needs the LlamaRisk report re-read with this question in mind. If the two are linked, the report's own Q1 headline case is really a Q2 own-token-collateral case. - **What the case gives Q5 that no other case gives (Inference).** - Dated debt-versus-token-depth observations for a governance token borrowed against inside its own ecosystem. - Concentration in one borrower. - Evidence that the binding resource was **OTC counterparties**, not on-chain depth. This is the nearest empirical anchor available for "debt ceiling relative to the collateral token's market depth", which C078 says cannot be calibrated. The CRV episodes do not calibrate a ceiling either. They do show what failure looks like when debt is a large multiple of on-chain depth, and G01 should name them. Other own-token or own-network candidates not examined (all *uncertain*, for the authors to check): - **Mento/Celo cUSD.** CELO in the reserve; the 2023 forum proposal "Returning CELO"; a commenter noting the CELO could not realistically be sold at prevailing liquidity. This is a survivorship counterweight: a stablecoin that kept a native-token reserve share and held its peg. [Celo forum](https://forum.celo.org/t/mento-reserve-returning-celo/5236). - **Venus VAI** (BNB Chain), with long periods of trading below peg. - **Waves USDN/XTN**, its 2023 continuation after the late-2022 rebrand. - **Kava USDX**, which was deprecated. ### 1.4 Q1 and Q4: Uniswap v4 hook coverage has survivorship bias The only hook in the report is StablePair (C025–C027). That is one launch-stage design, documented by its issuer, with no event study (C027 says so). The report omits the two 2025 hook-based losses: - **Cork Protocol, May 28, 2025.** *Fact (official post-mortem title/date and auditor analyses via search):* - 3,761 wstETH (~$11–12M) was taken from a wstETH:weETH vault of a **depeg-protection** product. - Root cause: missing access control on the hook's `beforeSwap` callback, so it could be called by something other than the PoolManager. This was combined with rate manipulation through an attacker-supplied exchange-rate provider. - The post-mortem says none of its audits flagged it. Sources: [Cork post-mortem](https://www.cork.tech/blog/post-mortem), [Dedaub](https://dedaub.com/blog/the-11m-cork-protocol-hack-a-critical-lesson-in-uniswap-v4-hook-security/), [BlockSec](https://blocksec.com/blog/cork-protocol-incident-two-independent-flaws-combine-into-one-devastating-exploit-chain). - **Bunni v2, Sept 2, 2025.** *Fact (auditor/press):* - ~$8.4M was drained on Ethereum and Unichain through rounding in a custom liquidity-distribution function inside the hook's swap path, compounded by repeated small withdrawals. - Bunni later shut down permanently, citing the cost of a secure relaunch. Sources: [BlockSec](https://blocksec.com/blog/bunni-incident-repeated-small-withdrawals-compound-a-rounding-error-into-an-8.4m-drain), [Halborn](https://www.halborn.com/blog/post/explained-the-bunni-hack-september-2025), [Decrypt](https://decrypt.co/345621/decentralized-exchange-bunni-pulls-the-plug-following-8-4m-flash-loan-exploit). *Inference:* imdUSD's peg hook is the component the brief singles out. These two cases show that the **hook's own code is a primary loss surface**, specifically callback authentication and rounding in custom pricing, and in Cork's case in a product whose purpose was depeg protection. C073 gives only economic advice for the hook and no security or invariant advice. A hook that takes on more responsibility for defending the peg also becomes a larger target. ### 1.5 Q4: fee evidence is thin and skips the obvious measured comparators C067–C069 rightly conclude that no clean surcharge study exists. The report could still have mined: - **Curve StableSwap off-peg fee multipliers during the March 2023 USDC depeg**, a natural experiment in pools with different multipliers. *Uncertain:* not checked. - **Maker/Sky PSM `tin`/`tout` fee changes**, which provide dated before/after volume shifts. These are custodial, but they are direct evidence of fee-driven venue migration. *Uncertain.* - **The earlier campaign's 2021 Fei "direct incentives" penalty.** The brief allows the R1–R6 campaign to be cited "where a later event revisits an earlier design". DeFiChain's 2024 discount fee (C065, C066) is exactly that kind of revisit, yet the report says only that 2020–21 is "outside this review". *Inference:* a single cross-reference would show that the surcharge has failed twice in different eras. That is much stronger evidence than one failure. --- ## 2. Survivorship and selection skew 1. **The survivors are well-capitalised and custodial; the failures are exotic.** *Inference.* - The cases that "survived" are crvUSD, GHO, frxUSD, Ethena and USDD. All rely on paired stablecoins, a GSM, custodied T-bills, $1.9B of stablecoin buffers or a USDT PSM (C002, C018, C023, C057, C043). - The failures are mostly off-chain-manager or privileged-mint cases (Stream, Elixir, USDX, Resolv). - Neither group resembles "small, overcollateralised by one thin volatile token, no custodial reserve". The groups that do resemble it (Liquity forks: Raft, Prisma, Felix; CRV-collateral lending) are missing. 2. **"Survived" for Ethena hides a large contraction.** *Fact (press citing CoinGecko/DIA, not issuer):* - USDe supply fell from ~$14.7B on Oct 9, 2025 to ~$6.4B by Dec 23, 2025 ([ForkLog](https://forklog.com/en/usde-capitalisation-by-ethena-plummets-by-8-3-billion-since-october/)). - A DIA page reports ~$3.9B by August 2026 ([DIA](https://www.diadata.org/map/global-stablecoins/stories/13-usde-synthetic-dollar-collapse/)). *Uncertain:* the August 2026 figure has not been checked against a primary series. - Binance reportedly reimbursed users over $283M for the October 2025 dislocation. C059–C061 describe primary redemption working, which it did. They omit that the liabilities shrank by roughly 55–75%. *Inference:* for imdUSD, a sustained slow run (redemptions grinding through low-rate troves for months) is a more realistic threat than a single 23-minute wick, and the report gives it no attention. 3. **The Djed/Indigo survival framing (C035–C040) leans on operator accounts.** C037 already flags this. The missing comparator is a native-collateral case that failed slowly (VAI or USDN) set against one that survived with a large reserve share (Mento). *Uncertain.* --- ## 3. Stale figures where newer numbers likely exist | Claim | Figure used | Problem | Fill | |---|---|---|---| | C058 | Ethena backing mix, **Aug 2025** (54% stable / 46% delta-neutral) | Predates both the Oct 2025 event and the supply contraction; the report concedes current docs are broader. | Ethena transparency dashboard / monthly governance update for Sept–Oct 2026 | | C059–C061 | Ethena redemption totals Oct 2025, and the 2024–25 events | No post-October supply path. | CoinGecko/DefiLlama USDe supply series, Oct 2025 → Oct 2026 | | C008 | crvUSD keeper ceilings May 2024 / Mar 2025 proposals | Acknowledged in G03, but no live on-chain read was attempted. The ceilings are a single public `debt_ceiling` read on the factory contract. | Etherscan read of the crvUSD ControllerFactory / Regulator at 2026-10-29 | | C017 | GHO peg stats for the 12 months to **Feb 2026** | Eight months old. GSM capacity is from Sept 2026 (C018), but price stats are not. | Same LlamaRisk methodology rerun, or a Dune query of the GHO/USDC pool | | C042–C043 | USDD API read of 2026-10-29 | Current, but there is no history. The Aug 2024 BTC removal (C041) rests on a secondary source. | Archived USDD reserve pages (Wayback) Jul–Sep 2024 | | C071 | Liquity v2 fee parameters from docs/repo | No check that the relaunched May 2025 deployment kept the same parameters. | Liquity v2 mainnet contracts (relaunched deployment) | --- ## 4. Self-reporting presented as if independent The report's convention (§ preamble) says "Fact … does not imply independent verification". That disclosure is correct. The trouble is in how the facts are then used: the **positive outcome findings rest disproportionately on issuer or retained-vendor data**, while the negative findings rest on third parties. *Inference.* - **C010 (crvUSD August 2026 "full monetary cycle", hourly range $0.99966–$1.00061)** comes from Curve's own monthly recap. It is the only positive crvUSD outcome measure. The only independent study, Pangea (C085), shows peg deviation getting **worse**. - **C017, C018 and C084** come from LlamaRisk and the Ethena risk committee. *Uncertain but checkable:* LlamaRisk is a contracted risk service provider to Aave and to Curve, and Ethena's risk committee members are engaged by the Ethena Foundation. These are informed, but not arm's-length. The report should disclose each analyst's engagement status. - **C057–C061** are Ethena's own post-mortem and committee material. C060 already shows internal inconsistency ($1.2B vs $1.9B). That should lower the evidential weight of the whole Ethena "survival" narrative, not only the one figure. - **C024 (Frax Facts 102.9% coverage), C036 (COTI flows) and C042–C043 (USDD API)** are issuer balance-sheet figures. The report labels them correctly, but §6 uses no independent counterweight for any of them. - **C046 and C054** come from Nexus Mutual, an insurer that paid claims on these events and is therefore an interested party. The 17-minute USR collapse figure should be reproduced from Curve pool logs. A table in the report listing, for each "survived" finding, the source's relationship to the protocol (issuer / retained provider / interested counterparty / independent) would fix this. --- ## 5. Stresses the design implications never model These are the main substantive gaps. Each names how the §6 recommendation could fail. All entries are *Inference* unless marked. 1. **Reflexive depth: L(s,T) is not independent of the stablecoin's stress (C076, C077).** - In an imdUSD-style system, redeemers receive staked IMD (or IMD) and must sell it into the same thin IMD market. - Liquidators and Stability Pool depositors receiving IMD do the same. - Borrowers deleveraging buy imdUSD, often by selling IMD. So every contraction path adds IMD supply to the very depth that C076 assumes is fixed. CRV in 2023 shows this: on-chain depth was effectively unusable, and the issue was resolved OTC. The formula needs L to be a function of the stress itself, or at minimum a requirement that L be measured *after* deducting the IMD that redemptions and liquidations would deliver into it. 2. **Thin-token oracle manipulation.** C070 and C071 assume redemptions pay "oracle-valued collateral". With a thin collateral token, the oracle is the attack surface. *Fact (auditor analyses):* **Polter Finance (Nov 16–19, 2024)** lost ~$8.7–12M when BOO spot reserves were drained on SpookySwap. An oracle that mixed Chainlink with pool reserves then over-valued 1 BOO deposited as collateral, against which ~9.13M wFTM was borrowed ([CertiK](https://www.certik.com/blog/polter-finance-incident-analysis), [Verichains](https://blog.verichains.io/p/polter-finance-attack-boo-token-price)). - **Oracle too high:** an attacker can mint against inflated IMD. - **Oracle too low or stale:** redeemers extract excess IMD from troves. *Inference:* imdUSD has a specific circularity. If the IMD price comes from a v4 pool the peg hook itself influences, the hook and the oracle can be manipulated in one transaction. The report never mentions oracle design for the collateral token. 3. **Staked-collateral liquidation mechanics.** C070 mentions unbonding delay once, for redemptions only. It does not model the following. - Liquidations or Stability Pool payouts in a staked token that cannot be sold until unbonding finishes. - Slashing. - A depeg of the liquid-staking wrapper against IMD, which is the Raft case: the bug sat in the collateral wrapper. The Raft failure is direct evidence that the wrapper layer needs its own supply invariants. 4. **Upward depeg with no mint-side inventory.** Without a PSM, the only above-peg defence is new borrowing. In a crash, borrowers need imdUSD to repay, so the stablecoin trades **above** $1 exactly when collateral is falling. crvUSD in June 2024 (C009) is the report's own example. - C073's hook advice is framed around defending below peg. - No recommendation covers what the hook does above peg when it holds no stablecoin inventory. - No recommendation covers the liquidations triggered when imdUSD-denominated lending markets read a premium price (the crvUSD spillover). 5. **Governance capture when the governance token is the collateral.** If IMD holders vote on collateral types, ceilings, oracle sources and hook parameters, an attacker who borrows or buys IMD can change the risk system that IMD backs. *Fact (widely reported, not re-checked here):* Compound's July 2024 "Humpy" proposal 289 passed by a concentrated voting bloc. That was not a stablecoin case, but it is the governance-token-capture template. *Inference:* timelocks, and parameter bounds that the borrowing token cannot vote around, are missing from §6. 6. **Concentration of a single borrower or LP.** LlamaLend in June 2024 had >90% of crvUSD borrows in one actor's accounts. The C072 and C076 caps are per-branch, not per-actor. A small stablecoin's early supply is typically concentrated in team, treasury or market-maker troves, and the report does not model that. 7. **Migration and upgrade risk.** Adding collateral types implies new branches, zaps and migration contracts. Prisma (migration zap) and Liquity v2's own 2025 redeployment both show the transition window as a distinct stress. C079 ("demonstrate the complete path at size") never mentions migrations. 8. **Slow run through redemptions.** Borrower-set rates mean that in a sustained below-peg period, low-rate troves are redeemed continuously. Their owners lose IMD exposure they never chose to sell, and supply shrinks (compare USDe −55% or more). The report never asks whether the system stays functional at a much lower supply: fixed costs, hook LP depth, Stability Pool size and minimum debt. --- ## 6. Recommendations that would fail under the stresses above | Recommendation | Fails when | Evidence the report should add | |---|---|---| | C070 "redemptions into collateral buyers can actually sell" | Redeemers dump IMD into the same thin market; the oracle lags; staked IMD cannot be sold during unbonding | CRV 2023 OTC resolution; Polter oracle manipulation; Raft wrapper bug | | C073 hook "improve trade economics… preserve corrective arbitrage" | The hook's own code is exploited (callback auth, rounding); the hook pool is also the price source | Cork (May 2025), Bunni (Sept 2025) | | C074 "keep safety gates from disabling the only useful response" | Correct, but the gates' governance can be captured by the collateral token | Governance-capture cases | | C076/C077 ceiling and solvency inequalities | L, h and s are correlated with the stablecoin's own stress and with a single dominant borrower | CRV June 2024 (LlamaLend concentration); USDe supply contraction for slow-run r | | C072 "cap each branch" | Concentration in one actor across branches; a shared router transmits shocks | LlamaLend June 2024 | | C079 "add per-period issuance bounds" | Correct. It should cite Yala as a second privileged-mint case and Raft as a collateral-wrapper mint case, so the bound must sit at the wrapper as well as the core | Yala YU, Raft R | --- ## 7. Suggested additions to `data.json` gaps These are proposed, not written into the dataset (out of my scope). - **G14 Liquity lineage record:** BOLD peg and redemption series since May 19, 2025; the Feb 2025 Stability Pool issue; Felix feUSD HYPE share and peg on 2025-10-10/11; Raft and Prisma post-mortems. - **G15 Own-ecosystem collateral in lending:** CRV debt versus on-chain CRV depth on 2023-07-31 and 2024-06-13; borrower concentration; OTC versus on-chain resolution. - **G16 Hook security record:** Cork and Bunni root causes mapped against the imdUSD hook's permissions and its custom pricing math. - **G17 Slow-run dynamics:** USDe supply path Oct 2025 → Oct 2026 from a primary or chain series; whether redemption-driven contraction impaired system function anywhere. - **G18 Source independence:** engagement status of LlamaRisk, Chaos Labs, the Ethena committee members and Nexus Mutual relative to each protocol they assessed. - **G19 Oracle design for thin collateral:** cases of collateral-oracle manipulation since 2023 (Polter and others) and their relevance to redemption pricing. --- ## 8. What the report does well (so the critique is not itself one-sided) - It does not over-claim on fees (C068, C069) or on ceilings (C078). - It keeps proposals, deployed values and issuer claims apart (C008, C011, C015). - It separates compensation from peg survival (C049, C055). - It flags its own inconsistent source (C060). The gaps above are about **which cases it examined**, not about the integrity of the ones it did. Even so, the selection matters for the decision: the missing cases (Liquity forks, CRV-collateral lending, v4 hook exploits, thin-token oracle manipulation) are the ones closest to imdUSD's architecture. ## 9. Limits of this critique - The leads in §§1–5 come from web searches run on 2026-10-29 and are mostly press or auditor summaries. Loss figures, lows and dates conflict between sources and are given as ranges. - I did not open the Cork or Prisma post-mortems in full, Felix's own documentation, or any on-chain data. - I did not verify the Compound proposal-289, Paxos PYUSD, Mento CELO-share, VAI, USDN or Nerite details. They are marked uncertain. - Citation accuracy of the reviewed report was out of scope and was not checked.