Aave has activated LlamaRisk agents for Pendle collateral on Plasma, the lending protocol confirmed. The agents can now adjust pricing and borrowing parameters for a principal token that currently has $36.72 million supplied. The move puts automated risk management in charge of a sizable chunk of Pendle-related exposure on the Plasma network.
What LlamaRisk agents actually control
The agents aren't free to do whatever they want. They operate under fixed step limits and cooldowns, meaning any parameter change is capped and can't happen continuously. That's a deliberate constraint — without it, an autonomous agent could reprice collateral faster than borrowers could react. The principal token in question has $36.72 million supplied. That's the pool the agents are now watching.
Pendle's principal tokens represent the principal portion of a yield-bearing asset, separated from its yield. They trade at a discount to the underlying asset's redemption value, and that discount narrows as maturity approaches. Pricing them correctly matters for Aave because they're used as collateral. Get the price wrong, and either borrowers get too much credit or liquidations fire when they shouldn't.
Why this is different from a normal risk parameter vote
Most Aave risk changes go through governance: a proposal, a vote, then execution. That process is slow by design. LlamaRisk agents change the timing. They can adjust pricing and borrowing parameters without waiting for a token vote, but only within the step limits and cooldowns that Aave set. The result is a middle ground — faster than governance, slower and more bounded than a fully autonomous system.
LlamaRisk has been one of Aave's risk service providers, so this isn't a new outside actor. It's an existing risk manager getting a new execution channel. The agents don't set the overall risk framework; they operate inside it.
The Plasma angle
Plasma is the network where this collateral sits. Aave's deployment there now has automated parameter control for this Pendle principal token. For borrowers, the practical effect is that pricing and borrowing parameters can shift between governance cycles. If the principal token's market price moves, the agents can respond without a proposal sitting in a forum for days.
That cuts both ways. Faster repricing can protect the protocol from bad debt if the collateral's value drops. It can also mean borrowers see their borrowing capacity change sooner than they'd expect from a governance-only system. The step limits are what keep that from being a sudden cliff.
What to watch
The $36.72 million supplied is the number that matters right now. If it grows, the agents are managing more risk with the same step limits and cooldowns. If it shrinks, the automated control covers less. Either way, the next parameter adjustment — whenever it comes and however large the allowed step is — will be the first real test of how this setup behaves in live conditions. There's no public schedule for when that adjustment might happen. The agents act when their conditions trigger, not on a calendar.
