Screen launchpad tokens by market cap, holders, top-10 %, curve and spam

Answers "which tokens match ALL of these criteria right now?" for launchpad tokens, either already bonded (stage=bonded: left the bonding curve for a regular pool) or still on the curve (stage=bonding). Omit stage when the question does not care which: both stages are screened into one list sorted by sortBy, stage echoes "all", every row says which it is in bonded and curvePct, and counts.byStage splits the counts. With no stage the liquidity floor and minLiquidityUsd apply to bonded rows only, and a row not bonded publishes liquidityUsd null: before bonding the figure counts the curve, not a tradeable pool, so it is neither shown nor checked. minCurvePct and maxCurvePct need stage=bonding; without it they are refused with a 400. Every criterion passed is checked against a value on the row, and every row carries the evidence it was checked on. A row is NEVER returned on a criterion that could not be checked: it is listed under unverified with the criteria it is missing instead. counts says how many candidates were read, checked, matched, dropped and left unverified. Market cap is read twice, from the candidate list and from the deepest pool, and the two can differ by a few percent; a market-cap bound passes only when every reading known passes it, so a row near a bound is left out rather than returned on the more favourable reading. A bonded row must also be physically possible: a pool holds at most twice the market cap in liquidity, so a bonded row listed with more than a set multiple of its market cap (2.2 by default) is re-read from its pools. Its liquidity becomes the pools' own total, and its market cap is kept only when the deepest pool confirms it within a set tolerance (10% by default); otherwise marketCapUsd is null, and the row is unverified with missing ["marketCapUsd"] whenever a market-cap bound or sortBy=marketCapUsd was asked for. excludeSpam=true keeps only rows whose spam is false: no mint authority, no freeze authority (on Solana the freeze authority is what makes a token unsellable) and not flagged rugged. Every row carries manipulationRisk: flagged with its reasons when the market cap of its pools outside the curve is far above their liquidity, nearly every trade is a buy, or supply is spread evenly across thousands of wallets. excludeManipulationRisk=true keeps only rows whose manipulationRisk.flagged is false. Bonded rows get a liquidity floor by default (echoed in appliedCriteria.minLiquidityUsd) because without one the bonded population in a market-cap band is dominated by farmed tokens with a few dollars of liquidity; pass minLiquidityUsd to change it, 0 to remove it. Results are sorted by sortBy (marketCapUsd, volume1hUsd or txns1h) in sortOrder, rows without the sort value last. The screen reads the leading candidates in that order, so a very narrow screen can return fewer rows than exist; unverified and counts show how much was read. ONLY SOL IS SERVED — every other chain 400s naming itself.