Salvagesalvagefi.com
Integration proposal — Aave

A priority queue for exits.

When a lending market reaches full utilization, withdrawals become first-come, first-served and the queue clears in a single block. Salvage lets a depositor who needs out now pay to go first, and lets one who is happy to wait get paid for waiting.

Aave's exits work. Four reserves are the exception.

We read all 358 reserves across Aave's 27 production markets, v2 and v3, on every chain it runs on. Four sit at 99% utilization or higher — one at exactly 100% — holding $2.8M that cannot be withdrawn. All four are legacy markets, frozen years ago. The most utilized live v3 reserve is at 92.9%, which is a reserve doing its job.

4
of 358 reserves at 99%+
$2.8M
supplied and unwithdrawable
92.9%
highest live v3 reserve
  • AMPL · Ethereum v299.45%$2,646,884sealed — the aToken has both transfers and burning disabled
  • USDC · zkSync v3100%+$99,834carries a deficit; $205 left in the pool
  • sUSD · Ethereum v2100.00%$13,243nothing left in the pool at all
  • renFIL · Ethereum v299.93%$7,796$5 left in the pool

Read on-chain 3 August 2026.

We checked these against real positions rather than a dashboard: a holder of 1,266,919 aAMPL cannot withdraw a single token. But that reserve is sealed at the contract level, so nobody can move those positions — us included. Strip it out and Aave's stuck balance is about $121,000, which is another way of saying Aave does not have this problem.

So this is not a fix for something that is broken. It is the exit already built for the day a reserve gets there.

At the top of the curve, exiting stops being a transaction and becomes a race.

A reserve at full utilization still lets people in and still pays interest. What it cannot do is pay everyone who wants to leave. Each scrap of returning liquidity is claimed in the block it appears, usually by a bot watching for exactly that, and a depositor refreshing the app never sees it. They did nothing wrong and there is nothing they can do differently.

  • The rate curve prices borrowing. It does not price exit priority — a depositor who needs out today and one content to wait a month stand in the same queue.
  • The scramble is invisible in the interface. A withdrawal that reverts reads as a broken app, not a market at capacity.
  • Depositors who get tired of it leave, and they leave when utilization is highest — exactly when the reserve can least afford it.

How it works

Nothing is added to Aave. No new contract, no governance vote, no change to the pool. aTokens are transferable ERC-20s and stay transferable when the pool cannot pay, which is the only property the design needs.

  1. 01

    The exiting depositor names their price

    They sign an order — their aTokens for an exact amount of the underlying, at a discount they choose. They see the figure before signing, and their tokens stay in their own wallet until it fills.

  2. 02

    The discount is the fee, and the market sets it

    Want out this hour while the reserve is tight, offer more. Content to wait, offer less, or do nothing at all and use Aave exactly as you do today. Nobody is opted into anything.

  3. 03

    Someone with patience takes the other side

    A buyer takes the position at that discount and holds it. They are buying yield below par from someone who wants out. Total supply is unchanged — the claim changed hands, it did not leave the pool.

  4. 04

    If nobody does, we fill it when liquidity returns

    We fund the payout, fill the order, redeem, and settle in one transaction. That path only runs when the pool can actually pay, so it never competes with ordinary withdrawals for scarce liquidity.

What Aave gets

  • Deposits stay in the pool. Selling a position leaves total supply untouched; withdrawing at the front of the queue takes liquidity out.
  • An exit that works under stress. The worst moment for a lending market is depositors finding out they cannot leave.
  • It is opt-in and entirely external. No integration risk, because nothing about the pool changes.
  • A live read on exit demand. What people will pay to leave is information the utilization curve does not carry.

What depositors get

  • A way out today at a price they set, instead of waiting on a queue they cannot see.
  • The choice to be on the other side — buy at a discount from someone who needs to leave, and earn the spread for holding through the crunch.
  • Their tokens stay in their wallet until the order fills, and they can cancel at any time without asking anyone.
  • An order that never fills costs nothing but a signature.

Why us

The hard part is not the order. It is landing the fill in the block liquidity appears, ahead of every other bot watching the same reserve. We run live MEV systems that compete for exactly that window, and the same infrastructure returned over $1,000,000 to depositors stuck in insolvent Euler vaults.

Terms

  • Building the integration is free. We do the work, at our cost.
  • We are paid out of the spread on a completed exit. No retainer, no listing fee, nothing invoiced to Aave.
  • If a depositor never exits, nobody pays anything.
Talk to us