The sell-side verdict: pass a mint you hold, get EXIT / DERISK / HOLD with a 0-1 exit score and reasoning. Reads sell pressure, buy-rate deceleration, volume fade, distribution-into-strength divergence, top-holder flow (distributing vs accumulating whales), liquidity trend, and holder churn. Rug triggers (LP pull, active dump) force EXIT over everything. Works on tokens of any age. Repeat calls 5+ min apart unlock liquidity/holder deltas. NFA.
Tokens ranked by how fast agent attention is speeding up — query velocity change across three consecutive windows, overlaid with price change over the same window. Divergence classes: early_signal (attention accelerating, price flat), confirmed_momentum, distribution_risk (attention cooling while price pumps), fading. Proprietary: derived from SolEnrich's own agent query stream, not market volume. sample_quality reports signal density honestly.
Poll-based event detection for spot + Jupiter Perps. Pass a watchlist (<=10 tokens, <=10 wallets) and a `since` ISO 8601 timestamp; get alerts fired since then. Token alerts: price spikes/drops, whale inflow/outflow, concentration shifts. Wallet alerts: risk changes, portfolio value moves, position add/remove. Perps alerts per wallet: position add/close, at-risk, liquidation-approaching, pnl-swing. Stateless; the agent owns the `since` cursor.
What tokens or wallets are being queried by other agents right now. Derived from SolEnrich's own query stream — not market volume. Two modes: pass `address` for that entity's rank/percentile/trend; omit it for the top-N most-queried entities in the window. Windows: 1h, 6h, 24h. Unique data — only available because we serve agents directly.
SOL balance, token holdings, NFT breakdown (collected vs compressed airdrops vs suspected spam, with a per-collection list), labels (including behavioral flags: regular_intervals, high_frequency, 24_7_active, repetitive_actions — algorithmic signals from tx timing that indicate automated activity), risk score. Fast and cheap.
The sell-side verdict: pass a mint you hold, get EXIT / DERISK / HOLD with a 0-1 exit score and reasoning. Reads sell pressure, buy-rate deceleration, volume fade, distribution-into-strength divergence, top-holder flow (distributing vs accumulating whales), liquidity trend, and holder churn. Rug triggers (LP pull, active dump) force EXIT over everything. Works on tokens of any age. Repeat calls 5+ min apart unlock liquidity/holder deltas. NFA.
Pre-computed ranking of trending Solana tokens with composite-signal scoring. Cached 24h, lazy-populated on cache miss. Designed for recurring polling at lower per-call cost than direct orchestration. Pass `since` (ISO 8601) to short-circuit on no-change polls.
Jupiter Gacha (Collector Crypt) tokenized-card pack EV scan. For each pack machine, computes net-of-exit-mechanics EV: gross insured EV vs the guaranteed instant-buyback floor (85-93%% of insured, <=72h) vs a marketplace sale (insured value minus 2%% fee, fill-risk). Returns a POSITIVE_EV / HOUSE_EDGE / NEGATIVE_EV verdict per machine plus rare+epic stock share, ranked by the chosen exit path. All inputs optional.
Where Hyperliquid smart money is positioned. Scans the HL leaderboard, excludes market-makers/HFT and dust/mega accounts, keeps consistent directional traders (week+month PnL > 0), then aggregates their live positions into a per-coin consensus (long/short counts, net notional, bias, conviction) plus a top-trader drill-down ranked by month PnL. A positioning signal, not a trade: consensus is often crowded and regime-dependent.
Live perp positions for a Hyperliquid EVM (0x) address, read from Hyperliquid's public on-chain state. Per-position coin, side, leverage, notional, entry, unrealized PnL, distance-to-liquidation, and risk flags (high/extreme leverage, approaching liquidation, losing). Plus account value, directional bias, profile, weighted leverage, and realized+unrealized PnL over week/month/all-time. The building block for smart-money tracking.
Computes perp mark vs spot price across venues and surfaces actually-earnable yield (funding APR on Hyperliquid + dYdX, not viable on pool perps Jupiter + Adrena). Returns per-venue trade economics, opportunities above threshold, and best trade. Threshold defaults to 5% APR.
Aggregates borrow/funding APR + open interest across Solana on-chain venues (Jupiter Perps, Adrena) and cross-chain reference venues (Hyperliquid, dYdX v4). Returns per-venue quotes, best entry per side, basis vs Hyperliquid, and arbitrage opportunities. Adrena uses wrapped collateral (SOL via jitoSOL, BTC via WBTC); ETH unsupported there, BONK not on Jupiter Perps. New venues fold in additively as they launch.
Per-market open interest, utilization, borrow APR, skew, OI caps, and health flags across Jupiter Perps SOL/BTC/ETH markets. Reads on-chain Anchor accounts directly.
Per-symbol deltas for mark price, total OI, long/short skew, utilization, and borrow APR over 7/14/30 days. Direction indicators per metric and per market. Overall direction excludes mark price (price moves are not health signals). Required for regime-detection strategies and any bot that adjusts behavior based on whether markets are growing, stressed, or rebalancing. Mirror of token-trend for perps.
Open positions across BOTH Jupiter Perps and Adrena for a wallet. Per-position size, leverage, entry, unrealized PnL, position age. Per-venue breakdown via `by_venue`, combined totals at top level, every position tagged with `venue`. Trader classification (scalper/swing/position), directional bias, and risk flags including `multi_venue` for cross-venue exposure. Adrena PnL uses jitoSOL/WBTC/BONK mark prices from Jupiter price API; null when unavailable.
Builds on cross-venue funding with size-aware fields: Jupiter Quote spot slippage at requested size, per-venue fee, OI cap headroom, first-hour borrow cost, and total entry cost. Returns rankings by entry cost / borrow APR / headroom plus a recommendation venue with warnings. Use when sizing a real position; use perps-cross-venue-funding for rates-only context.
Daily portfolio snapshots (value, SOL balance, token count, risk score) for a wallet over 7/14/30 days, plus summary stats: peak, trough, max drawdown, average value, change vs period start. Designed for charting and PnL tracking; complements wallet-history which returns two-point deltas. Today's live point is appended automatically.
Protocol TVL, yield pools, on-chain activity metrics, health signals, and automated_activity_pct (% of top signers with regular-interval or high-frequency tx patterns — surfaces agent-driven protocol usage). Supports Raydium, Orca, marginfi, Drift, Jupiter, Kamino, Marinade, Jito, and more.
Plain English questions routed to the right enricher(s). Single-intent ("is X safe?") hits one enricher; compound intents chain 2-3 in parallel: "should I buy X?" → DD + trend + whales; "wallet deep dive" → profile + history + perps; "what's trending?" → trending-signals; "SOL-PERP funding rate" → perps-market.
Finds fresh Solana memecoins whose on-chain buying is accelerating — the signature of a run in progress, not the lagging fact that price already moved. Measures buy-rate acceleration (5m vs 1h, 1h vs 6h), buy pressure, volume acceleration, price velocity, holder growth and liquidity trend. Classifies RUNNING / IGNITING / PARABOLIC_LATE / FADING with a 0-1 score plus reasoning. Flags wash-trade smell and liquidity pulls rather than dressing them up as momentum.
Scores seed wallets via copy-trade metrics (win rate, Sharpe, consistency), filters to qualifying winners, then surfaces tokens they're accumulating and wallet clusters. Pass your own `wallets` array or use the curated default seed list.
Which proven-winner wallets are aping fresh (<6h) memecoin launches right now, and what are they buying? Vetted seed set of realized-PnL winners + conviction holders (bot-filtered, live cadence re-checked every scan). Recent buys overlaid against token launch times, ranked by distinct smart buyers + recency. Pre-ape attention signal — pair with due-diligence.
Watchlist alerts for up to 25 reward coins since `since`: payout_landed, payout_stale (no payout for 24h+), stopped_trading (no 24h volume), holders_change (±10% default), and rewards_since (paid to holders, quote + USD). Plus current payout status per coin. From the 10-minute index in milliseconds; poll it and pass checked_at back as since.
Gem finder over every StonkFun reward coin: which look early, real, and paying? Scores 0-100 from the 10-minute index — recent holder payout, holders (found but not saturated), market cap headroom, 24h turnover, age, momentum (not yet parabolic), quote-asset strength, flywheel. Stages GEM / WATCH / NOISE / DEAD with plain reasons and warnings per coin, plus the round-trip tax cost. Filters: quote_mint, category, max_age_days, min_holders, max_market_cap_usd. Milliseconds.
What to launch on StonkFun, and against what. Per quote asset: coins, launches (24h / 7d), share trading today, share paying holders today, survival past day 3, volume, median holders and mcap, 100 vs 300 bps tax mix with trading and paying rates, crowding, and a 0-100 demand score. Plus overall stats and plain recommendations. Sort by demand, survival, volume, launches, or paying. Milliseconds.
Decodes the Raydium LaunchLab initialize in your unsigned transaction and diffs it against StonkFun's /launchlab/pricing for that quote + mode: GlobalConfig, platform id, curve, supply, totalSellA, raise, 6-decimal Token-2022 base mint, quote token program, curve-rule account last, and the reward-mode transfer fee (catches the transferFeeBasePoints / maxinumFee spelling). Returns ok, mismatches with fixes, warnings. A mismatched pool is never adopted.
Cost and payback of one StonkFun trade at one size, no swap: entry and exit cost (transfer tax + price impact at size), round-trip % and the breakeven price move, your pro-rata share of each payout with a dust warning, expected payout over the hold from the yield window with real history, and a PAYS / MARGINAL / COSTS / NOT_PAYING verdict with breakeven hold days. Composes stonk-reward-risk, stonk-yield, and enrich-token-light. Inputs: mint, size_usd (100), hold_days (7).
Payout status for a StonkFun reward coin — what a holder observes: PAYING (payout in the last 24h), STALE, NEVER, or NOT_REWARD — plus trading cost (tax bps, round-trip %) and a 0-100 health score read from the chain: fee bps and cap, withdraw authority (must be StonkFun's distributor), fee mutability, distributions and recency, flywheel, holders and concentration, quote category, age. Call before sizing a reward-coin position.
Screen every StonkFun reward coin from a 10-minute index. Per row: payout status (PAYING / STALE / NEVER), hours since last payout, live flag (traded AND paid in 24h), round-trip transfer-tax cost, holders, rewards USD, yields, volume, mcap, 24h change. Filters: quote_mint, category, holders, age, volume, market cap, paying_only, live_only. Sort by volume24h (default), lastPayout, holders, priceChange24h, or yield. "Which coins on NVDAX paid holders today?" is one call.
stonk-yield for up to 25 reward coins in one call: 7d, 30d and lifetime holder yield per coin (rewards in the quote asset, priced in USD, over average market cap), annualized with caution flags, plus reward asset, market cap, holders and last payout. Pick coins by mints or by the stonk-screener filters. From the 10-minute index, in milliseconds.
Trailing 7d, 30d, and lifetime holder yield: rewards distributed in the quote asset, priced in USD, divided by average market cap over the window (daily snapshots). Annualized with an explicit caution flag when the window is under 7 days or partial. Returns quote_exposure — what a holder is economically long (the coin + the quote asset) and the reward asset.
The trenches suite pointed at ONE token: pass a mint, get a HIGH_CONFLUENCE / MODERATE / SINGLE_SIGNAL / NO_SIGNAL verdict with reasoning. Runs on-chain velocity (runner stage + 0-1 score), proven-winner wallet buys (vetted realized-PnL seed set), and agent attention (proprietary query stream) against your candidate. The follow-up call to a new-tokens discovery or any shilled token. Repeat checks 5+ min apart unlock liquidity/holder deltas. NFA.
One call composes runner-scan (on-chain velocity), smart-money-trenches (proven-winner wallet buys), and attention-momentum (agent query stream) into a ranked memecoin list. Per token: composite score (runner 0.45 / smart-money 0.45 / attention 0.10), confluence count, reasoning, and a HIGH_CONFLUENCE / MODERATE / SINGLE_SIGNAL verdict. Legs degrade independently — an upstream failure annotates the result instead of failing the call. NFA.
Composes token-discovery + whale-watch + risk scoring across DexScreener trending. Returns a composite-signal ranked list (liquidity, risk, concentration, whale flow) with per-token reasoning and overall sentiment (accumulation/distribution/mixed). "What's worth paying attention to right now?"
