com.getplexa/mcp

Two-level economic-safety check for an ERC-20 on Base, Polygon or Arbitrum before buying it. LEVEL 1 — verdict: "avoid" ONLY when a listed trap is PROVEN on-chain at that block (NO_EXIT_VENUE: no pool exists to sell back into; EXIT_LIQUIDITY_DRAINED: the counter-asset pot is smaller than 5% of the size you asked about — the pot is an upper bound on what a sale can return, so under 5% you recover at most 5 cents on the dollar; TRADING_DISABLED: the token's own trading gate reads false). "clear" means none of them fired — it is NOT a safety rating and NOT a prediction, so do not treat it as "safe". LEVEL 2 — risk_profile: token age, holder concentration (ANY token age: top-1/top-10 share of supply computed from holder balances and on-chain totalSupply; the data age travels in the response, and a vendor outage answers "unknown" with an explicit reason, never a clean default), liquidity structure including exitLiquidityUsd (the dollars a sell can actually reach, across up to TWO hops: pairs against WETH/USDC directly, plus pairs against any other counter asset whose own exit we confirmed — see EXIT MEASURABILITY below), oracle availability, and fee-on-transfer / blacklist / limit fingerprints — reported as measured DATA with plain-language notes, not as severities. Weigh them yourself: on 40 outcome-labelled Base tokens those structural factors marked live and dead tokens at the same rate, which is why they no longer produce a verdict. CONFIDENCE means different things on the two branches: on "avoid" it is confidence in the TRIGGER, which is a measurement taken that call, so it is always "high" — do NOT gate on it (`if confidence < high: ignore` would skip a proven trap); on "clear" it is how complete the checks were, and there it is worth reading. Gaps are always reported explicitly in `reasons` and as nulls in `risk_profile`. FRESHNESS AND COMPLETENESS: `validUntil` (epoch ms) is until when asking again returns THIS answer — a statement about our recomputation, not a…

Get EXECUTABLE on-chain liquidity for a swap (Base / Polygon / Arbitrum) before trading: the real fill price under a given size (not the mid/spot quote), price impact in bps, realizable depth, the route (per-leg dex/pool/fee), worst-case slippage, and a confidence rating. Sources Uniswap v3, Aerodrome, Aerodrome Slipstream (concentrated liquidity) and Uniswap v4 via canonical on-chain quoters — v4 is SIMULATED per pool through the canonical V4Quoter so hooks execute; a pool whose hook rejects the simulation is left out of the route rather than guessed at. On Polygon and Arbitrum the quoters are Uniswap v3 and Algebra (QuickSwap, Camelot). Call this when an agent needs to know what price a swap of a specific size will ACTUALLY execute at, especially for thin/long-tail tokens. `spotPrice` is NULLABLE: when no reference mid is defensible at this size, `spotPrice` and `priceImpactBps` come back null rather than confidently wrong — an impact measured against a moving origin is worse than no impact at all. Every response carries `validUntil` (epoch ms): asking again before it returns the same answer. It is a statement about when WE recompute, not a promise that the market held still. Cost: $0.02 USDC per call via x402 (paid automatically by the configured wallet). FIRST CALL FROM A NEW WALLET IS FREE: the full response, every field, same engine as a paid call — only the COUNT is limited (one per wallet, shared across both endpoints). Sign the x402 payment as usual; nothing is settled on-chain for that first call. You do not have to take that on trust: every 200 response carries `freeTrialApplied` — true when THAT call was served free, false when it was billed. MARKET CROSS-CHECK: every quote also carries `marketRoute` {source, amountOut, impactBps, splitTop[{venue, share}]} — what an aggregator offers for the SAME input in the same minute — plus `divergenceBps` (how far OUR output is from it; negative means we give less) and `singleRouteOnly`. `singleRouteOnly: true` mea…