On-chain data reveals a peculiar pattern: the Sanaa Protocol's governance votes have a 99.8% correlation with a single address labeled 'TehranVault'. Over the past seven days, the protocol's total value locked (TVL) dropped 40% as liquidity providers fled, yet the governance continued to pass proposals that aligned with TehranVault's voting history. The ledger remembers what the interface forgets.
Sanaa Protocol presents itself as a decentralized lending and borrowing platform, complete with a DAO, yield farming, and a flash loan feature. It claims to be community-run, with no single point of failure. But a forensic audit of its smart contracts tells a different story. The protocol's core architecture mirrors a proxy war: a local entity (the Houthi-like front) appears autonomous, but the strategic decisions—interest rate models, liquidation thresholds, oracle selection—are hardcoded to a master key. This key, in the contract SanaaController.sol at line 247, allows the owner to override any governance decision with a single transaction. The interface shows a DAO, but the code reveals a dictatorship.
During my audit of Ethereum 2.0's slasher protocol in 2017, I learned that consensus failures often hide in seemingly redundant checks. Here, the redundancy is a facade. The setLiquidationThreshold function at line 312 is gated by onlyOwner, not onlyGovernance. The owner address is immutable and set during deployment to a multisig wallet—but all five signers of that multisig are controlled by the same entity, TehranVault. This is not decentralization; it is a centrally planned economy dressed in Solidity. The protocol's whitepaper boasts of 'resistance to censorship,' but the code ensures resistance to community control.
Let us examine the oracle integration. Sanaa uses a custom price feed that aggregates data from three sources: Chainlink, Uniswap V3 TWAP, and a proprietary 'Sanaa Oracle.' The catch: the Sanaa Oracle is a simple contract that returns whatever the owner sets. At line 487, setSanaaPrice(bytes32 asset, uint256 price) external onlyOwner allows the admin to override the aggregated price. This is the exact mechanism that Three Arrows Capital used to manipulate their isolated margin positions—I analyzed that forensic trail in 2022. The result: a single entity can trigger liquidation cascades at will, extracting value from unsuspecting depositors. The ledger remembers what the interface forgets.
The contrarian blind spot here is the assumption that multisig governance equates to security. Many auditors look at the number of signers and assume distributed control. But if all signers are part of the same organization—or worse, the same person using different wallets—the multisig is a placebo. Sanaa's five signers all originate from the same IP range and were funded by a single wallet at block 15,234,567. This is not a secret; it is a pattern that any blockchain analyst can detect. Yet the protocol's marketing team continues to emphasize 'community governance' while the code explicitly grants veto power to TehranVault.
Another critical oversight is the protocol's 'autonomous AI agent' for liquidations. The agent is supposed to run on-chain and execute liquidations based on market conditions. But the contract's executeAgentAction function at line 612 requires a signature from a predefined 'oracle operator'—again, TehranVault. The AI is a puppet. During my work on the AI agent payment layer specification in 2026, I insisted on zero-knowledge proofs to ensure agent autonomy. Here, the agent has no autonomy; it is a remote control. The code does not lie; auditors just listen.
What are the strategic implications? Sanaa Protocol is not a rogue project; it is a deliberate instrument of a larger entity—call it 'TehranVault.' The protocol's purpose is to drain liquidity from other DeFi ecosystems, much like the Houthis' role in the Red Sea is to disrupt global shipping for Iran's benefit. The 'war' is economic: Sanaa's flash loans and price manipulation are the missiles; the depositors are the shipping lanes. The protocol has already extracted over $50 million in arbitrage profits, all funneled back to TehranVault. The ledger remembers what the interface forgets.
The takeaway is a vulnerability forecast: protocols with hidden master keys will face a catastrophic liquidation event when the master decides to exit. The slasher does not forgive. Neither do we. As I wrote in my MakerDAO CDP analysis, the system's redundancy held because of conservative parameters. Here, the parameters are weaponized. Expect a cascade within six months—not from external attackers, but from the protocol's own creators. The question is not if, but when the proxy war escalates into open exploitation.