Magic Internet Money Depegged to Near $0.14. Its Collateral Is On-Chain and Readable. That Was Not Enough.

On July 3, 2026, Magic Internet Money trades near fourteen cents. That is not a typo and it is not a thin-liquidity wick that snaps back in an hour. MIM, the crypto-collateralized dollar Abracadabra.money has run since 2021, has been sitting far below its dollar peg for more than a week, and the main pool that is supposed to hold it at a dollar has had its good side almost entirely drained. A read straight off Ethereum puts MIM near $0.1433 on the largest Curve pool and near $0.1478 against Tether on Uniswap, roughly 86 percent below a dollar. Two independent trackers agree within a fraction of a cent.

This piece is not a victory lap and it is not a hack post-mortem, because there was no hack. It is a careful read of the single most instructive property of the MIM break: MIM's collateral is on-chain. You can read it. There is no custodian, no off-chain fund, no attestor standing between you and the backing. And it depegged anyway. If you take one idea from this series, take this one, because MIM is the cleanest proof of it we have: being able to verify the reserves yourself is necessary, and it is not the same as the reserves being sound. Verifiability bounds who you have to trust. It does not eliminate collateral risk, liquidity risk, or the risk that a peg simply breaks.

What actually happened, and when

The collapse was gradual and it was public the whole way down. By mid-June the peg was already slipping: on June 15, 2026, MIM was trading around $0.82, roughly 18 percent below a dollar, and Abracadabra injected about $100,000 into the primary MIM Curve pool to steady it. On June 18 the protocol went further, announcing a liquidity incentive program distributing about 140 million SPELL, its governance token, to try to deepen pool liquidity. Neither held. On June 24 MIM plunged to roughly $0.49 to $0.50, about half its peg and a drop of around 36 percent over 24 hours by on-chain trackers, on thin, one-sided liquidity that met a broad market selloff.

That was the point Abracadabra declared emergency measures. In its own framing, it began to raise interest rates across every single cauldron, including markets it had already flagged as deprecated, and it suspended direct incentives and Curve bribes until MIM returns to its peg. The logic it stated is worth quoting because it tells you exactly what kind of fix this is: the depeg, the team said, "creates a natural incentive for borrowers to repay debt at a discount, accelerating supply contraction and strengthening the path back to the peg." In plain terms, make borrowing expensive so borrowers buy back cheap MIM to close their loans, shrinking supply until the price recovers. It is a supply-side maneuver, not a demonstration that the backing is whole.

The maneuver has not, so far, restored the peg. The June coverage was written with MIM near $0.50; the on-chain reality on July 3 is near $0.14. The supply-contraction plan is visibly working in one narrow sense, which we come back to below, and it has not brought MIM back to a dollar. Two clarifications so this is fair. First, this was not an exploit. Abracadabra has been hacked before, in separate and unrelated incidents, a January 2024 cauldron bug and another in late 2025, but the June 2026 depeg was not reported as a hack by anyone covering it, and nothing on-chain suggests one. Second, MIM is not paused or wound down; it is a live token in an acute crisis, which is a different and in some ways harder thing to read than a clean shutdown.

The half you can read yourself, in two commands

Start with supply, because it is the one number no one can spin. MIM lives at 0x99D8a9C45b2ecA8864373A26D1459e3Dff1e17F3 on Ethereum, and its total supply on July 3, 2026 is about 174.9 million, read with a single call anyone can run: cast call 0x99D8a9C45b2ecA8864373A26D1459e3Dff1e17F3 "totalSupply()(uint256)" --rpc-url https://ethereum-rpc.publicnode.com. That figure is the total minted, and it is larger than the amount actually in circulation, because a CDP stablecoin keeps a large reservoir of unborrowed MIM inside its own contracts ready to be lent. DefiLlama, which nets that out, tracked only about $4.1 million of MIM as circulating on the same day. Sit with that for a second: the borrowed float, the MIM that is actually out in the world as somebody's debt, has collapsed to a few million dollars. That is the supply-contraction plan working, borrowers repaying under rate pressure, and it is also why the price is so fragile: a very thin float meeting almost no bid.

Now the pool, because the price is set there. The main Curve pool that pairs MIM against the 3pool stablecoins sits at 0x5a6A4D54456819380173272A5E8E9B9904BdF41B, and on July 3 it held roughly 4.28 million MIM against only about 12,900 units of 3Crv. A healthy stable pool is close to balanced. This one is almost entirely MIM, which is what a completed run looks like: everyone who wanted out swapped their MIM for the good stablecoins on the other side until the good side was gone, and what remains is a pool that is nearly all MIM quoting a price near fourteen cents. You do not need us for any of this. It is a block explorer and one cast call. The standing, always-live version of this read is at kerne.fi/verify/mim, and you can run the identical check on any other token with the free tool at /verify-anything.

Why readable collateral still lost the peg

Here is the uncomfortable part for anyone who thinks on-chain collateral is the end of the diligence conversation. MIM is minted as debt against collateral deposited in Abracadabra's cauldrons, which are isolated on-chain lending markets. That collateral is readable in the cauldron contracts. There is no attestor to trust on the reserve leg, which on our verify pages is the strong end of the axis. And none of that stopped the peg from breaking. The trust boundary for a CDP dollar was never "do you trust an attestor." It is "does the on-chain collateral actually cover the outstanding debt at prices it can be liquidated into, and is there enough two-sided liquidity to defend the peg when confidence goes." Those are questions the chain can inform but does not answer on its own, and in June the market answered them the hard way.

Put MIM next to the year's other breaks and the shape is familiar. Where the failure hid off-chain, holders found out late: Stream's loss sat with an off-chain manager, Resolv's mint authority sat behind a single off-chain key, which is why we audited our own key surface in public. Where the backing was legible, holders could watch the trouble arrive in advance rather than discover a hole afterward: apxUSD traded below par in full view as its preferred-share collateral fell, which we walked in our apxUSD piece, and Synthetix's sUSD, an on-chain, over-collateralized dollar, still lost its peg for reasons you could read on-chain, which we walked in the sUSD wind-down. MIM belongs firmly in that second group. The move was painful, but for anyone actually reading the chain it was not a surprise, and Abracadabra's own fix, raising rates to shrink supply, is a statement that this is a debt-and-liquidity problem to be worked down, not a stolen-reserves problem to be recovered. That distinction matters a great deal if MIM is in your book.

If MIM is in your book

If you are a treasury, a fund, or a DAO that held or lent against MIM, the depeg is not just a price on a screen. It is a number you have to put in front of your own LPs, your auditors, or your token holders: what was this exposure worth, and as of when. The team whose token broke is not the right source for that number, and their published response is a rate schedule, not a proof of backing. What is defensible is a third-party, point-in-time, cryptographically signed read of exactly what a given address held on-chain at a given block, a statement anyone you forward it to can re-derive themselves in three lines without trusting you or us.

That is a thing Kerne sells, on the same signing stack it runs over its own reserves every hour. The entry rung is a self-serve, machine-signed read of any single public address at a block, $29, paid in USDC on Base and signed on the page in about two minutes, at /instant-read. A human-reviewed version of the same signed read of one address is $149 at /address-read. And for an allocator who needs a scoped, independent read of a counterparty's on-chain reserves and the authority around them, a proof you hold rather than the issuer's dashboard, that is a fixed $2,500 single-counterparty read, or a $5,000 to $15,000 review for a portfolio or several chains, at /counterparty-verification. Every one of these signs objective on-chain bytes at a point in time. None of them is a solvency opinion, a valuation, or an audit, and we say so on each page.

Where we stand, so this is not an abstraction

A piece that points at a depegged competitor and exempts its author would be the exact thing this series warns against, so, plainly. kUSD is structurally different from MIM in the way that matters here: it mints one-to-one from USDC through an on-chain module on Base, so the reserve is readable directly off the chain, and it publishes an hourly Proof of Reserves signed with an EIP-191 key that you re-derive yourself against public inputs at /verify, with no attestor in the path. But MIM is a reminder that structural verifiability is a floor, not a guarantee, so the boundary gets named in the same breath. kUSD's delta-neutral hedge leg runs on Hyperliquid, a single venue, self-reported and signature-bound rather than independently re-derivable, an independent attestation of it is being scoped, and Kerne is pre-audit at Genesis scale with a live modeled yield and a near-empty insurance fund. We did not remove every risk. We moved the trust boundary to the smallest surface we could and named it, and we publish the open gaps at /legible rather than wait to be asked.

The check to run on any CDP dollar

MIM leaves a short, portable checklist for the next crypto-collateralized dollar. One: can you read the collateral on-chain at all, and if you can, do not stop there, because MIM could and still broke. Two: what is the actually-circulating float versus the total minted, and how thin is the pool that sets the price, both of which you can read yourself today. Three: when the peg is under stress, is the issuer's response a proof that the backing is whole, or a plan to shrink supply until the price recovers, because those are very different promises. The free tool at /verify-anything runs the on-chain half of that walk on any token, and the synthetic-dollar scorecard ranks the field on exactly this axis. Verifiable was never the same as safe. MIM is what it looks like when a token you could fully read on-chain breaks anyway.

Figures are as of July 3, 2026 and move intraday; nothing here is investment advice or an offer of any token. MIM's contract address, its total supply of about 174.9 million, and the main Curve MIM/3Crv pool balances of roughly 4.28 million MIM against about 12,900 3Crv are direct on-chain reads on July 3, 2026, reproducible with the commands above and viewable at Etherscan; the price near $0.1433 to $0.1478 and the roughly $4.1 million tracked as circulating are per DexScreener and DefiLlama the same day. The June 15 level near $0.82 with a $100,000 Curve injection, the June 18 distribution of about 140 million SPELL, the June 24 fall to roughly $0.49 to $0.50 (about 36 percent over 24 hours), the emergency measures raising interest rates across every cauldron and suspending direct incentives and Curve bribes, and Abracadabra's stated logic that the depeg incentivizes borrowers to repay at a discount, are per Crypto Briefing, Cointelegraph, and crypto.news (June 24 to 25, 2026); the depeg was not reported as a hack, and MIM's earlier, unrelated exploits in January 2024 and late 2025 are referenced only to keep them separate from this event. Kerne is not affiliated with Abracadabra. Kerne's own claims resolve to live endpoints: the hourly signed Proof of Reserves at /api/por/signed, the reserve breakdown at /api/por, and the live risk surface at /api/risk-status. A /verify pass proves an attestation is authentic and fresh; it is not an audit and not a solvency opinion.