Security1 distinct publisher2 min readPublished
The Moonwell post-mortem puts externally supplied capital at $1.95m. Recycled through repeated borrowing, that money moved $7.5m of MAMO buying. That ratio, rather than the loss total, sets an attacker's entry price.
The Watch · Security desk

invest
An attacker burned $3.8M in MAMO slippage to borrow $10M of Moonwell depositors' assets1 distinct publisher
invest
Ajna's oracle-free design loses $775,000 to its own liquidation arithmetic1 distinct publisher
invest
A 2022 oracle hack starts moving again, three days after Pando shut its books1 distinct publisher
build
All 575 listed x402 services split $516.96 over thirty days1 distinct publisher
Compiled by The WatchSomething wrong?How this is made
The donation step is the one to study. Formally supplying 15,089,595 MAMO bought mMAMO receipts worth about $159,904 at the reference price of $0.010597 [4][7]. The next 53,393,290 MAMO arrived as a plain transfer into the mMAMO contract, roughly 3.5 times the size of the supplied position, and because it minted no shares the market's mint-path supply cap never registered it [5][9]. Every outstanding share got heavier instead. The principal account held about three quarters of outstanding mMAMO, so it captured most of that increase, and its underlying claim grew to about 55.51 million MAMO [8].
The buying did the rest, a 40.7 times move in the MAMO/USD source feed [4]. At the peak the enlarged claim marked at roughly $23.94 million, about 150 times the baseline value of the position it started from, and $11.97 million of borrowing power at the market's 50 percent collateral factor [9][5]. Moonwell's highest archived accepted price was lower, $0.40248571, worth $22.34 million and $11.17 million of capacity [10]. The borrows stopped at 98.7 percent of that lower ceiling [6]. The actor sized borrows against Moonwell's accepted price, using the feed only as the lever that pushed collateral value up.
The cost figures in the post-mortem contradict its own headline number. About 799 ETH became 1,947,391 USDC and crossed to Base [2]. Gross MAMO purchasing volume of roughly $7.50 million against that principal is 3.85 turns of the same money, because borrowed assets were spent again as purchasing power [3][1]. Gross borrowings came to 5.66 times the outside capital [2]. The traced stablecoin increase, which Moonwell labels a gain estimate rather than a settled figure, is $6.785 million before gas, 3.48 times the money that entered [15][3].
The exit ran through one corridor: 8,729,454 USDC burned on Base via Wormhole's CCTPv2-with-Executor helper over Circle CCTP V2, 8,728,319 USDC received on Ethereum, a gap of 1,135 USDC, then conversion through Velora into canonical DAI and transfer to the linked operational account [12][8]. Evidence is cut at Base block 50,524,962, 14:01:11 UTC on 27 August, with supplementary verification through 28 August [16].
About $9.131 million of borrower obligations remained at that cutoff on Moonwell's oracle basis, and the post-mortem presents it as a best estimate of residual debt and potential bad debt rather than a closed number [14].
Ranked by verification strength, evidence, and original report placement.
On 27 August 2026 an actor exploited Moonwell's MAMO market on Base by combining collateral-accounting inflation with manipulation of MAMO's oracle price.
The operation was seeded with approximately 799 ETH, which a linked funding wallet converted into 1,947,391 USDC and bridged to Base.
The wallets used approximately $7.50 million of gross purchasing volume to acquire MAMO, much of which consisted of borrowed assets recycled during the operation; Moonwell states the clearest measure of externally supplied principal is the initial $1.947 million USDC.
The principal account accumulated MAMO and formally supplied 15,089,595 MAMO to Moonwell, receiving mMAMO receipt tokens.
The account transferred another 53,393,290 MAMO directly into the mMAMO contract without minting new mMAMO, increasing the MAMO backing each existing share and raising the exchange rate by approximately 3.68 times without consuming the market's mint-path supply cap.
The linked accounts purchased approximately 94.31 million MAMO through decentralized exchanges while liquidity was thin, and the MAMO/USD source feed rose from approximately $0.010597 to $0.43127363.
Distinct publishers with included, body-backed reporting in this cluster.
1 article · August 30, 2026
Follow any of these and your For You feed starts watching them — no settings page required.
Evidence-backed comparisons of source perspectives and observed adoption signals. Read the methodology
Which Builder, Operator, and Investor concerns the observed source mix emphasized—not a truth score.
Evidence, demonstrated adoption, hype gap, incentives, and confidence are assessed independently, each on its own current evidence. How these are measured.
Forensic detail, single narrator
Addresses, block heights, event counts and per-block oracle valuations are all here, down to 595 liquidation events across 594 transactions and a 1,135 USDC discrepancy across the bridge. That is not the texture of a guess. What it lacks is a second pair of eyes: the chain is public, but nobody outside Moonwell has reproduced any of it in this reporting, and the two most consequential dollar figures are labelled estimates by the author itself.
Live market, real assets out the door
Nothing here is a testnet demonstration. MAMO had been a governance-listed Moonwell market since October 2025, cbBTC, WETH, USDC and wstETH actually left the pool, third-party liquidators competed for the collateral within 32 seconds, and the proceeds crossed to Ethereum on Circle's production CCTP path. The roughly $9.1m still owed is the measure of how much genuine capital was standing behind the accounting.
Argues down its own biggest number
Most incident write-ups reach for the largest figure available. This one does the opposite: it refuses to call the $7.5m of MAMO buying outside capital, names $1.947m as the honest principal, hedges both the $6.785m gain and the $9.131m shortfall, and separates the raw feed peak from the price the protocol actually accepted. If anything it undersells the transferable finding — a plain token transfer walked around the supply cap that governance had set as the market's risk limit, and that sentence deserved more than a bullet point.
Only the losing party is talking
Moonwell controls the framing of an incident that cost Moonwell's users money. That shapes what is foregrounded and what is not: gross borrowings are reported without netting liquidations, the shortfall is dated to 'current oracle prices' that will drift, MIP-B48 and the original listing parameters get a single line, and remediation goes unmentioned. The disclosure is detailed and self-critical in places, which cuts against the worst reading — but no party with an opposing interest has checked a figure in it.
Trust the mechanism, date the totals
Two different reliabilities are bundled here. The causal chain — donate into the receipt-token contract, inflate the exchange rate, buy up a thin book, borrow against the reprice, get liquidated 32 seconds later — is documented tightly enough that we would treat it as established. The dollar totals are softer: they are frozen at one Base block, computed on the protocol's own oracle, and unverified by anyone else, so the shortfall in particular should be read as of 28 August 2026 and no later.