Vault Agent monitoring
Portfolio management does not end once funds are allocated. Markets move, yields change, liquidity fluctuates, incentives expire, and new opportunities appear. Instead of making a single decision at deposit and walking away, the Allocation Engine continuously reassesses both your existing positions and the alternatives available to them.
Monitoring is continuous, but changes are only proposed when they would improve portfolio quality or safety — not on every pass.
What’s monitored
Section titled “What’s monitored”Each monitoring pass evaluates the supported protocols, your current positions, and the candidate opportunities around them, including:
- Liquidity — TVL, withdrawable liquidity, and available deposit/lending capacity for each position.
- Yield — interest rates, vault share price or exchange rate, and whether a position’s yield is still competitive against other approved opportunities.
- Rewards — pending and accruing reward emissions, and whether claiming or converting them is currently worthwhile.
- Collateral and asset quality — collateral composition and underlying asset quality for lending-style vaults; a previously eligible asset can become ineligible if its on-chain liquidity, trading volume, or market depth deteriorates.
- Protocol health and eligibility — whether the protocol, vault, and adapter are still registered, unpaused, and meet the current safety requirements.
- Market conditions — asset prices, rates, and reward-token values that affect a position’s real return.
- Portfolio exposure — concentration, diversification, and your profile or preference limits across the whole agent, not just one position.
These inputs are evaluated together with your chosen profile and preferences before any change is proposed.
Verify, don’t trust
Section titled “Verify, don’t trust”Wherever practical, YieldSeeker reads protocol state directly from on-chain contracts — TVL, available liquidity, deposit capacity, vault accounting, lending rates, and share prices — rather than relying solely on third-party dashboards or protocol-reported statistics. This reduces dependence on external services and keeps allocation decisions grounded in the current on-chain state of each supported protocol. See Design principles.
Protocol safety parameters
Section titled “Protocol safety parameters”Not every monitoring input is something you or your profile can adjust. Alongside market conditions and your own preferences, the platform maintains a small set of safety parameters that apply across every agent and cannot be overridden by an individual user.
If a supported vault or protocol is identified as presenting an unacceptable risk — a security incident, an operational issue, or another material concern — it can be marked ineligible for future allocation platform-wide, and affected agents may have their positions flagged for exit as part of a coordinated response. This lets the platform react quickly to a specific incident without changing how monitoring works for every other supported opportunity. See Governance, registries, pauses, and timelocks for the underlying registry and pause controls.
When the agent acts
Section titled “When the agent acts”A monitoring pass finding a problem, or a better opportunity, does not execute anything on its own — it feeds the same Allocation Engine that evaluated your original allocation. A rebalance, exit, or reward claim/conversion is only proposed when the expected benefit justifies the transaction cost and stays within your profile, preferences, and portfolio constraints; monitoring alone is not a promise of action. See Vault Agent decision pipeline and Vault Agent vault selection for how a candidate change is actually chosen.