Skip to content
YieldSeeker Docs

Borrow Agent monitoring

Borrow monitoring watches a market price against a health-factor boundary and makes the agent eligible for its next run. MARKET PRICECOLLATERALMONITORINGHEALTH FACTORRule boundaryELIGIBLE EARLYNext Autoseek run

Monitoring continuously checks measurable Borrow Agent rules against the market data for positions the agent holds. It watches risk boundaries; Autoseek decides what to do when it runs, and the Execution Framework carries out only permitted actions.

For each open borrow position, the health-factor rule creates warning and breach boundaries based on the collateral price that would reach those health-factor levels. Monitoring also checks the market’s collateral-price feed and whether the market accepts new borrows. If you set a collateral-price floor, that adds another boundary.

The agent overview shows whether a boundary is clear, approaching, or breached. Borrow risk alerts can report a health-factor warning or breach, a collateral-price floor, a market that has stopped accepting new borrows, an oracle problem that prevents the position’s risk from being verified, or a risk signal on the market’s collateral or borrowed token.

YieldSeeker also continuously watches every borrow market agents use or can use, and the collateral and borrowed tokens behind them: prices, token supply and DEX liquidity. While any of these signals a problem, no agent borrows more or adds collateral in the affected markets; open positions stay, and a market you have a position in shows a warning in the agent’s monitoring status and sends a risk alert. If a market, a protocol or one of its tokens presents an unacceptable risk, such as a security incident, YieldSeeker can open a risk incident for it. Agents in an affected market are woken to move their loan to another market that passes their rules or, when none does, to close the position: the agent repays the loan with idle borrowed tokens, sells collateral to cover any shortfall and keeps the remaining collateral idle in its Agent Wallet.

A triggered boundary can make the agent eligible for its next Autoseek run early. Monitoring does not execute a transaction or choose a market move. At the next run, Autoseek rechecks current prices, debt, rules, market status, and available liquidity before it plans an action. Enabling a risk-alert channel is required for the agent to borrow, but it does not prevent liquidation.

For the default deleverage response, the agent uses idle borrowed tokens to repay toward its target when health factor falls below the warning level. Below its minimum level, it can sell collateral, swap the proceeds for the borrowed asset, and repay toward the target. Other risk responses can limit this action; see Borrow Agent preferences.

A lending protocol can liquidate a position when its health factor reaches 1.0 or below. Price changes, borrow interest, market liquidity, oracle conditions, transaction execution, or a delay before the next run can leave the agent unable to act in time. Monitoring and automated deleveraging do not guarantee that a position will avoid liquidation.

See Risks and limitations before funding an agent.