Aave Risk Team Proposes a “Rate Limiter” That Would Slow Borrowing During Active Exploits

MakeCryippreferred on

• LlamaRisk’s proposal introduces a dynamic borrow-cap buffer, starting at 10% of outstanding debt with a $1 million floor and a $25 million ceiling, that refills gradually as normal borrowing occurs
• Modeled against an actual $123 million rsETH-related draw, the new buffer would stretch that withdrawal to roughly two hours instead of the minutes it originally took
• A separate October 2026 Binance-related $44.6 million draw was modeled under the same system and would take between 31 and 36 minutes, rather than occurring near-instantly

A governance proposal from risk-management firm LlamaRisk, published to Aave’s forum on November 11, 2026, introduces what the proposal calls a dynamic caps oracle rate limiter, a mechanism designed to slow down abnormally large, rapid borrowing from Aave’s markets without blocking legitimate activity under normal conditions. The core idea is a borrow-cap buffer that starts at 10% of a market’s outstanding debt, bounded by a floor of $1 million and a ceiling of $25 million, and then refills gradually over time as ordinary borrowing activity occurs.

Under normal market conditions, a buffer that refills steadily as users borrow and repay would be largely invisible to typical activity, since legitimate borrowing rarely approaches the pace needed to exhaust it. The mechanism is specifically aimed at scenarios where an attacker, having compromised an oracle, a bridge, or some other upstream system, attempts to drain a market through an unusually large, rapid sequence of borrows, a pattern that has characterized several major DeFi exploits in recent years.

LlamaRisk’s proposal backs up its design with concrete historical modeling. Run against an actual $123 million rsETH-related draw from a prior incident, the firm’s simulation shows the new rate limiter would have stretched that withdrawal out to roughly two hours, compared with the mere minutes the original draw took to execute under Aave’s current, unthrottled borrow-cap structure. A separate test case, modeled against an October 2026 incident involving a $44.6 million draw connected to Binance, shows the mechanism would have extended that withdrawal to between 31 and 36 minutes.

The proposal describes this rate limiter as the first stage of a broader initiative LlamaRisk calls its “Risk Layer,” suggesting further related risk-management mechanisms may follow in subsequent governance proposals. By converting what would otherwise be a near-instantaneous drain into a process stretched across tens of minutes to a couple of hours, the mechanism is explicitly designed to buy response time, giving Aave’s own risk team, white-hat researchers, or automated monitoring systems a meaningful window to detect and potentially intervene in an attack before it fully completes.

The proposal represents a notable evolution in how major DeFi lending protocols think about exploit defense, moving beyond purely reactive measures like post-incident freezes and toward proactive, built-in friction designed to slow an attack while it is still unfolding. The full proposal is published on Aave’s governance forum.