Borrow Agent preferences
Choose a starting profile, then set a borrow amount and supported risk preferences. The Allocation Engine uses the effective rules to choose whether and how much to borrow; market availability and platform safety controls still apply.
Starting profiles
Section titled “Starting profiles”The Borrow Agent offers two profiles:
| Profile | Minimum health factor | Warning health factor | Target health factor | Maximum borrow APY | Minimum available liquidity |
|---|---|---|---|---|---|
| Conservative | 1.25 | 1.50 | 1.80 | 8% | $1,000,000 |
| Explorative (default) | 1.10 | 1.30 | 1.50 | 12% | $100,000 |
Both profiles start with a borrow amount of zero, use the deleverage risk response, and allow Aave V3, Aave V4, and Morpho Blue. A health-factor rule must keep its minimum at or above 1.05 and below its warning level, which must remain below its target.
When the agent borrows more
Section titled “When the agent borrows more”When the collateral price rises, the agent borrows more once the health factor is as far above the target as the warning level is below it. With a 1.5 warning and a 1.8 target, that is 2.1. Depositing more collateral normally lets it borrow straight away.
Two preferences slow this down:
- Borrow-more health factor: for example, “only borrow more once my health factor reaches 2.4”. It must be above the target. New collateral is still supplied, but the agent borrows more only once the health factor reaches that level. A fixed borrow amount you set still applies.
- Time between borrows: for example, “wait at least 30 days between borrows”. The agent takes no new debt until that many days have passed since its last borrow. It still supplies collateral, repays, deleverages, closes and moves loans between markets in that time.
The borrow-more health factor doesn’t delay a loan on a position with no debt. The waiting time counts from the agent’s last borrow, so it doesn’t delay its first one.
Borrow amount and risk response
Section titled “Borrow amount and risk response”The borrow amount is the total debt the agent aims to keep, not an additional amount on top of the current debt. You can set a fixed amount in the borrowed asset, a USD value, or ask the agent to borrow as much as its safety rules allow. The agent does not borrow until you set a nonzero amount.
Risk responses govern only health-factor breaches. Below the warning level, the agent uses idle borrowed tokens to repay toward the target. deleverage may sell collateral to cover a shortfall below the minimum; repay_only does not sell collateral for a health-factor response, so liquidation remains possible if idle funds are insufficient. alert_only does not repay because of health factor alone. Other rules can still trigger repayment regardless of the response: a crossed collateral price floor, or a YieldSeeker risk incident on your market with no other market to move to, closes the position, selling collateral as needed; borrow APY above its maximum or debt above the borrow amount can trigger repayment from idle borrowed tokens.
Other supported preferences
Section titled “Other supported preferences”Active memories can set or refine rules for:
- minimum, warning, and target health factors
- an optional top-up health factor above the target, which gates additional borrowing after idle collateral is supplied
- permitted lending providers and minimum market liquidity
- the maximum borrow APY
- a collateral reserve to keep idle in the Agent Wallet
- a minimum collateral price that closes the position if crossed, selling collateral as needed to repay debt regardless of the risk response
- whether the agent may sell collateral in response to health-factor risk
- a minimum number of days between successful borrows
A memory that replaces a rule type must describe the complete desired settings for that type. Review the effective rules in the app after changing a preference. A rule change makes the agent eligible for an Autoseek run straight away; it does not guarantee a transaction if the market or position fails its checks.
Simulate before saving
Section titled “Simulate before saving”You can simulate draft memories to review the rules they produce and the position the agent would leave under current conditions. A simulation is a projection; it does not save the memories or execute a transaction.