Skip to main content
Every pool your Agent can use has been reviewed by us first, and it stays under watch for as long as there’s money in it. The work happens in three steps. Before a pool goes live, our risk team looks at the protocol, the collateral behind it and the pool itself, and decides which strategy tier it belongs to. Once it’s live, its numbers are refreshed throughout the day and checked against the limits of that tier. When something goes wrong, the engine doesn’t wait for its next scheduled run. Risk events such as a depeg, an oracle gap or an exploit in progress are sent to it the moment they’re detected, and it stops new deposits or pulls the funds out. How a pool gets in, how it’s watched, and how risk reaches the engine.

Strategy Tiers

What Conservative, Aggressive and Yield Maxxing each allow.

Whitelisting

What a protocol, a collateral and a pool need to get in.

Ongoing Monitoring

What we watch once a pool is live, and when we stop or exit.

Incident Response

What happens when something breaks.
Every live pool’s rating, checks and alerts are on the Risk Dashboard.

Strategy Tiers

Every pool sits in one of three tiers. The tier tells you how much risk you’re taking, and it comes mostly from one question: what is the pool actually exposed to?

Conservative

5–8% target return. For capital you want to keep safe first.

Aggressive

8–14% target return. More yield, with riskier collateral allowed.

Yield Maxxing

10–18% target return. Highest yield, with slower withdrawals on some pools.
Targets are native returns without incentives, averaged across market cycles over roughly three years.

Exposure caps

We look through each pool to the collateral behind it, and measure how much of the pool’s exposure sits in collateral of each tier. A pool gets the first tier whose caps it meets: Risky means collateral that fails even the Yield Maxxing criteria. Each collateral’s own tier comes from Collateral Whitelisting.
Exposures of 0.5% or less are ignored as dust. If less than 60% of a pool’s exposure has been reviewed, we don’t guess: the pool is marked as having insufficient data.

Also required

Conservative pools never use leverage or looping. Exposure moves as curators reallocate, so these caps are checked again every night. See Pool Monitoring. For how strategies are picked and customized, see Yield Strategies.

Whitelisting

Nothing reaches your Agent without passing three reviews, in this order. A pool is only as good as the weakest of the three.

Protocol Whitelisting

We start with the protocol, because everything else is built on its code and its team.

Collateral Whitelisting

Lending pools are only as safe as the collateral borrowers post. Each collateral is reviewed on its own, using CoinGecko, Chainlink and Pyth feeds, GeckoTerminal, DexScreener and the issuer’s documentation. A collateral gets a tier only if it passes every line for that tier: It also needs enough backing for its category:
A depeg only counts if it lasts at least 8 hours with at least $5K of trading volume, so a short wick on a thin market doesn’t disqualify an asset. If data is missing on two or more lines, the asset is marked as having insufficient data instead of being given a tier.

Pool Whitelisting

New pools are reviewed every day. Some rules apply to every pool, and vaults and lending markets each have their own on top.

Ongoing Monitoring

Getting whitelisted is the start, not the end. Markets change, curators reallocate, and a pool that was fine last week can turn. So every live pool is checked in two ways:
  • On a schedule: pool metrics are refreshed through the day, and every collateral and pool is re-rated each night.
  • On events: a depeg, an oracle gap or an exploit in progress is sent to the engine as soon as it’s detected, without waiting for the next run.
Each check can do one of two things. It can stop new money going in while existing positions stay put, or it can move existing funds out. See also our Security Agents & Human Team and Emergency Response.

What can go wrong

Collateral Monitoring

  • Daily re-rating: every day, each collateral’s backing and every rule of its tier are checked again from fresh data. If the result no longer matches its tier, the risk team is alerted.
  • Depegs: base assets and collateral stablecoins are watched for peg breaks as they happen.
  • New collateral: when a curator adds a collateral to a vault we use, it’s flagged and reviewed against the collateral criteria.

Pool Monitoring

These checks run on every live pool, whatever its type.Pool performance history

Pool Rating

Every live pool gets a simple score on the Risk Dashboard: the number of checks it passes out of the checks that apply to it. Risk Dashboard filters

Incident Response

When something breaks, the engine acts first and people review after. Automatic actions protect funds in the minutes that matter. The risk team then looks at what happened and decides whether a rating should change. Ratings are only ever changed by a person.