Survey staking and lending opportunities on Base: APY, TVL, project and asset for each. Use this to see the landscape or to answer 'what is available', and use best_yield instead when one asset's single best option is wanted. Results already exclude obvious traps (minimum TVL, APY ceiling). Returns market data, not a recommendation; the pool id of any row can be passed to yield_history.

The single highest-APY pool on Base for one asset, with junk-farm filters applied (minimum TVL, APY ceiling, no impermanent-loss exposure by default). Use it for a direct 'where is the best rate for X' question. It ranks by rate alone, so before committing capital check market_profile for what the curator charges and yield_history for whether the rate has held.

The Base lending shelf an agent can actually act on: Morpho ERC-4626 vaults, Aave V3 reserves and Moonwell markets, each with the contract address to deposit into (depositTo) and the asset's address and decimals. Morpho entries also carry a risk block: curator fee, vault age, allocation concentration, and how much of the APY comes from incentives rather than borrower interest. Aave and Moonwell rates are read from their own contracts, so markets missing from aggregator snapshots appear here.

Daily APY and TVL history for one market, with min, median, max and volatility over the window. Answers the question a live rate cannot: whether today's APY is typical for this market or a spike that appeared yesterday. Call it before acting on any unusually high rate. observations says how many daily snapshots exist — a market added recently will have few, and a short series should not be read as stability.

Risk profile for one Morpho vault or Aave reserve, for use before committing capital. Returns score (0-100) and band ('solid', 'fair', 'thin' or 'fragile'), plus a signals object holding the six inputs behind it: size (TVL), track_record (days running), stability (APY volatility from daily snapshots), yield_quality (share of the APY that is borrower interest rather than incentives that can stop), cost (the curator's percentage cut) and diversification (share sitting in the largest position). Each signal carries its raw value and its sub-score, and method.weights gives the weighting, so the score can be recomputed. A signal lacking data reports score null and is excluded from the weighting rather than guessed at. Pair with yield_history for an unusually high rate. Describes what has happened, not what will.