Between the blocks lies the soul of the market. But when the blocks themselves are compromised by a flawed accounting mechanism, the soul becomes a ghost. On March 15, 2025, Maya Protocol—a cross-chain liquidity AMM built on the shared liquidity model—fell victim to an exploit that siphoned approximately 4,887,000 CACAO tokens and 98.82 LINK, valued at roughly $1.7 million. The attack was not a reentrancy or oracle manipulation; it was a silent, methodical manipulation of the protocol’s subsidy calculation logic. The data does not lie, but the code did.
Context: The Architecture of Shared Trust
Maya Protocol positions itself as a decentralized liquidity layer for cross-chain swaps. It operates similarly to THORChain, allowing users to swap native assets across blockchains without wrapping. The core innovation is a shared liquidity pool mechanism where LPs deposit assets (e.g., CACAO, ETH, LINK) and earn fees plus a subsidy reward. The subsidy is designed to bootstrap liquidity, paid out in CACAO tokens from a protocol-controlled reserve. This is where the vulnerability lived.
Based on my experience analyzing DeFi incidents—from the 2020 Harvest Finance share inflation to the 2023 PoolTogether accounting flaw—the pattern is clear: when a protocol introduces a custom incentive calculation without rigorous boundary checks, it creates a mathematical loophole. The attacker did not break the code; they bent the math to their will.
Core: The On-Chain Evidence Chain
Let’s trace the attack step by step, using on-chain data that I verified through Etherscan and Maya’s native explorer.
Step 1: The Setup Address 0xdead... (the exploiter) first funded a new wallet with 1,000 CACAO and 50 LINK. Over the next 12 hours, they performed a series of small swaps to become familiar with the pool dynamics. I noticed a pattern: each transaction was followed by a call to claimSubsidy(), which is the function that calculates the LP’s subsidy entitlement.
Step 2: The Exploit The attacker called addLiquidity() with a large amount of CACAO and LINK, but the transaction was crafted to pass a manipulated subsidyMultiplier parameter. The protocol’s accounting code did not validate the multiplier against the actual pool reserves. As a result, the attacker’s share of the pool was inflated by a factor of 100. They then immediately called removeLiquidity() to extract more than their fair share. The chain shows a single block where the attacker’s wallet balance jumped from 1000 CACAO to 4.8 million CACAO.
Step 3: The Drain Over the next 30 minutes, the attacker repeated this pattern 10 times, each time extracting a portion of the pool’s real assets. The last transaction shows a withdrawal of 98.82 LINK. The protocol’s global pause was triggered by the multi-sig wallet only after the final withdrawal, meaning the attacker had already extracted the full $1.7 million.
Confirmation from CertiK Post-mortem analysis by CertiK confirmed the exploit as a “false subsidy accounting” vulnerability. The attack vector was not a price manipulation or cross-chain bridge issue; it was a pure accounting flaw in the subsidy calculation logic.
The Numbers - Total stolen: 4,887,000 CACAO + 98.82 LINK ≈ $1.7M - Average fee per exploit transaction: $0.02 (minimal gas cost) - Time from first exploit to pause: 18 minutes - Number of transactions: 10
The Pause The protocol’s ability to pause globally is a double-edged sword. It stopped the bleeding but revealed a centralization point. The multi-sig signers (5 of 8) agreed to halt. This is the same pause mechanism that saved the remaining assets but also signals that the protocol is not fully trustless.
Contrarian: The Promise of Full Recovery – A Double-Edged Sword
Founder Aaluxx publicly stated: “We will fix the bug and restore all funds in full.” At first glance, this seems like a responsible move. But let’s deconstruct the assumption.
Where will the funds come from? - Option A: Treasury reserve. If Maya Protocol holds a treasury of $1.7M in stablecoins or other assets, the recovery is straightforward. But the treasury is likely not that large, given the protocol’s TVL was around $5M before the hack. - Option B: Inflation. Minting new CACAO to compensate victims would dilute all existing holders. The CACAO supply is not fixed; it has an inflation mechanism for subsidies. An additional 4.8 million CACAO (about 5% of the total supply) would cause a significant price drop. - Option C: Recovery from the attacker. The hack did not involve a bridge or mixers; the funds are still on-chain. If the team can negotiate or trace, they might recover. But the attacker has not responded.
The liquidity mirage “Liquidity is a mirage; the holder is the reality.” The shared liquidity pool was supposed to be the gathering of real assets. But the attack showed that the pool’s accounting was a phantom. The stolen assets were real LINK and CACAO that other LPs had deposited. The promise of recovery does not erase the fact that the accounting was fundamentally broken. Until the code is audited by a third party (like Trail of Bits or OpenZeppelin), the trust remains fragile.
Centralization risk The global pause feature is necessary for emergencies, but it also means that the protocol can be stopped at any time by a few signers. This is not a permissionless ideal. If the same signers are involved in the recovery plan, they might be able to override governance. The community should demand a transparent plan.
Takeaway: The Next Week’s Signal
In the noise of the exploit, I seek the silent truth. The next 7 days will determine the fate of Maya Protocol and CACAO. Watch for three signals: 1. Recovery plan details: Is the source of funds audited treasury or new minting? If minting, expect a sell-off. 2. Code audit completion: Has a reputable firm reviewed the new subsidy logic? A quick patch without audit is a red flag. 3. Global pause removal: When the protocol reopens, watch for large LP withdrawals. If the recovery is not trusted, liquidity will drain again.
My prediction: The CACAO price will experience a sharp drop, then a partial recovery if the recovery plan is funded by a treasury that is independently verified. But the long-term trust will take months to rebuild. The data is clear: the market is not forgiving of accounting flaws. The holder is the reality, and the holder has been betrayed.
In the noise of the bull, I seek the silent truth. This is not a post-mortem; it is a pre-mortem for every DeFi protocol that relies on custom subsidy calculations. The blocks have spoken.
