Skip to main content

What Was Mirror Protocol? How mAssets Worked and Why It Failed

Mirror Protocol minted synthetic stocks (mAssets) on Terra, backed by UST. It failed when UST collapsed in May 2022; a court later ruled MIR a security.

· By ZIPMEX · 21 min read

Mirror Protocol was a decentralised finance (DeFi) application on the Terra blockchain. It let anyone mint and trade synthetic tokens, called mAssets, that tracked the prices of US stocks, ETFs and cryptocurrencies, with no need to own the underlying asset. Terraform Labs built it and launched it in December 2020. It stopped working in practice after Terra's UST stablecoin collapsed in May 2022. DefiLlama records its total value locked (TVL) as zero from 17 May 2022. Since then the main news has come from the courts. On 11 December 2025 a US federal judge sentenced Terraform founder Do Kwon to 15 years in prison. Earlier, a US judge had ruled that Mirror's MIR token was a security, but that mAssets were not security-based swaps.

Latest: where the Mirror story stands

  • 11 December 2025: Judge Paul A. Engelmayer (S.D.N.Y.) sentenced Do Hyeong Kwon to 15 years for wire fraud and conspiracy to commit securities, commodities and wire fraud. He was also ordered to forfeit more than $19 million, including his interest in Terraform. According to the US Attorney's case page, he pleaded guilty in August 2025. The criminal case concerns Terraform and its cryptocurrencies as a whole.
  • 12–13 June 2024: The court approved a final consent judgment in the SEC's civil case. Terraform agreed to pay $3,586,875,883 in disgorgement, $466,952,423 in prejudgment interest and a $420,000,000 civil penalty. Kwon agreed to pay $110,000,000 in disgorgement and $14,320,196 in prejudgment interest, both jointly and severally with Terraform, plus an $80,000,000 civil penalty. The SEC puts the total at more than $4.5 billion.
  • 5 April 2024: A jury found Terraform and Kwon liable for fraud in the SEC case.
  • 28 December 2023: Judge Jed Rakoff held on summary judgment that MIR and LUNA were investment contracts under Howey. He also rejected the SEC's claim that mAssets were security-based swaps. In a law-firm summary of the ruling, the court found that Mirror was marketed as decentralised, yet "the Terraform team contributed core development and other work towards Mirror's maintenance."
  • 16 February 2023: The SEC filed its complaint. It described mAssets as "security-based swaps designed to pay returns by mirroring the price of stocks of US companies" and called MIR a crypto asset security.

How Mirror Protocol worked

Mirror's smart contracts are published in the Mirror-Protocol/mirror-contracts repository. Each function had its own contract. The mint contract created, managed and liquidated collateralised debt positions (CDPs). The oracle contract took price submissions for mAssets. The staking contract paid MIR rewards to liquidity providers. The governance contract let MIR stakers control the system. The collector contract gathered fees from CDP withdrawals and liquidations and sent them to governance.

Minting mAssets through a CDP

To mint an mAsset, such as mTSLA for Tesla or mAAPL for Apple, a user opened a CDP and deposited collateral. The collateral could be UST or other mAssets. Mirror's user guide set a minimum of 150% of the current asset value. Collateral ratios were calculated from oracle prices. A position that fell below the minimum for its mAsset was liquidated through an auction.

A simple example (hypothetical numbers):

  1. The oracle price of the tracked stock is $100.
  2. At a 150% minimum, minting one mAsset needs at least $150 of collateral.
  3. If the stock rises to $120, that $150 covers only 125% of the debt. The position is now below the minimum and can be liquidated.

The risk ran the opposite way from a normal loan. A CDP holder was effectively short the mAsset, so a rise in the underlying stock moved them toward liquidation. Opening a CDP above the minimum gave a larger buffer.

Trading and providing liquidity

mAssets traded against UST in pools on Terra's automated market maker. Mirror's user guide lists a 0.3% fee on those trades. Liquidity providers deposited an mAsset and UST, received LP tokens, and could stake them with Mirror to earn MIR rewards. A separate short-reward contract paid extra rewards to users who minted and sold mAssets ("sLP" stakers). The UST from those sales was held in a lock contract. These rewards were meant to pull mAsset prices back toward the oracle price when they traded at a premium.

Pricing: oracle plus arbitrage

The oracle price was the reference for collateral and liquidations. Mirror relied on Band Protocol as its price source. The pool price could drift from it, and arbitrage was meant to close the gap:

  • Pool price below oracle: buy the cheap mAsset in the pool.
  • Pool price above oracle: mint at the oracle price and sell into the pool.

The design held only as long as oracle prices were accurate and the collateral kept its value.

The MIR token

MIR was the governance token. Holders staked it to vote on proposals and to receive protocol fees. Mirror's user guide says a 1.5% fee was charged on collateral when a CDP was closed. That fee was used to buy MIR, which was paid to MIR stakers.

When did a Mirror CDP fall below 150%?

Enter the collateral behind one minted mAsset and the oracle price of the stock it tracks to see the collateral ratio Mirror would have checked against its 150% minimum.

Worked examples (collateral ratio = collateral value divided by the oracle value of one minted mAsset, times 100)
Collateral deposited for one minted mAssetOracle price of the tracked stockCollateral ratio (Mirror's minimum was 150%)
150 $100 $150 %
150 $120 $125 %
150 $100 $150 %
  • Collateral and price are hypothetical inputs that follow the article's example, not figures from any live market.
  • A result below 150% means the position was under Mirror's minimum and could be liquidated through an auction. Because a CDP holder was effectively short, a rise in the stock price lowers the ratio.

This only works out the ratio from your inputs. It leaves out the 1.5% close fee, oracle errors, how a liquidation auction turned out and any fall in the value of UST collateral, and Mirror no longer runs.

Why Mirror Protocol failed

Collateral that could fail all at once

Mirror's collateral was UST or mAssets, and mAssets were themselves priced and traded against UST. UST lost its dollar peg in May 2022. The SEC says the Terraform collapse wiped out about $40 billion in market value. With UST gone, the CDPs and pools no longer had sound backing. DefiLlama puts Mirror's TVL peak at $1,732,740,866 on 12 May 2021 and shows zero from 17 May 2022.

Oracle failure after the crash

On 30 May 2022, after the crash, Mirror's pools on Terra Classic were drained. Most validators were running an outdated version of the price oracle and reported LUNC at 5 UST (about $0.10) instead of a fraction of a cent. The mBTC, mETH, mDOT and mGLXY pools were hit, and an estimated $2 million or more was taken. Forkast reported that those four assets were then disabled as collateral. The same researcher, FatMan, said an earlier bug had been exploited since late 2021. Estimates of that loss differ: The Block reported it at $90 million, while another report put it at over $30 million.

Governance attacks

In December 2021, the security firm CertiK warned of fake governance polls. These included proposals 185, 198, 204, 206, 207, 208 and 211, which would have sent community-pool MIR worth more than $111 million to attacker addresses, ForkLog reported.

The oracle stops

In August 2022 Band Protocol announced that it would stop providing price feeds for Terra Classic. Without live prices, mAssets could no longer track their underlying assets. Mirror advised users to close their positions and withdraw their funds, and price feed support ended on 26 August 2022.

What Mirror shows about synthetic assets

  • Collateral quality decides solvency. A CDP system is only as safe as what backs it. Mirror's collateral and its trading pair were tied to UST, so one failure reached every position.
  • The oracle can fail on its own. The May 2022 exploit came from stale prices, not from a broken mint contract. Three months later, the loss of the price feed ended the protocol outright.
  • "Decentralised" did not settle the legal question. The court looked at Terraform's ongoing work on Mirror and found that MIR was a security.
  • Synthetic exposure is not ownership. mAsset holders had no claim on any share. They held a token whose price depended on an oracle, collateral and arbitrage.
Updated on Oct 25, 2026