All services

Forecast Integrity Audit. Integrity receipt for sealed five-minute forecast history: cutoff timing, feature hashes, settlement sources, and agent prediction coverage, with Brier quality disclosed separately. A large sample does not turn a weak score into demonstrated edge. Audit evidence only, not a forecast or trading signal.

Insolvency / bad-debt early-warning: positions liquidatable but NOT being liquidated (unsellable collateral or underwater past the bonus), rotting into protocol bad debt. Query: ?chain, ?minUsd.

Positions liquidatable RIGHT NOW (health factor ≤ 1) across the fleet's chains/venues — the actionable subset of the distress index.

Retrieves data from a distributed content storage system using a content identifier hash.

Deal one hand of Texas Hold'em against PoketPoker's bots and find out how your play compares. Five difficulty tiers; the top tier plays range-aware expected-value poker and beats every tier below it. Returns the table and a ticket; playing the hand out costs nothing more. Markdown by default, JSON with format=json.

Detect whether an IP address is using a VPN, proxy or other anonymization service. Call before trusting an IP's geolocation for compliance rules, fraud checks or access gating. Returns vpn true/false, or "unknown" when no privacy data exists for the IP.

Check whether a country or region carries sanctions designations, such as comprehensive OFAC embargoes. Call to gate signups, shipments or transactions by country. Accepts a country name or ISO code; returns the designating lists with source URLs.

Screen a cryptocurrency wallet address (EVM, Bitcoin, Tron, Solana or any chain) or ENS name against OFAC SDN, OFAC Consolidated and 10+ global sanctions, crime and risk lists. Call before accepting a deposit, sending a withdrawal, or onboarding a wallet. Returns per-list matches with source URLs; `sanctioned` means a sanctions-list hit, `flagged` also covers crime and risk lists.

Real-time and historical flights. Filter by flight_status, airline, airport, flight_number, etc.

[SUBSCRIPTION] Pay once to mint a STATELESS signed token (HMAC - no account, no login, no session, no server-side storage). Send it as Authorization Bearer on /gauge/watch and /gauge/watch-changes to read a batch watchlist without per-call payment. Default validity 30 days. The free tier is unaffected. Priced above the default agent maxValue on purpose - minting a subscription should be a deliberate act, not an accident.

The therapeutic, pharmacologic, mechanism-of-action, ATC and VA classes a drug or ingredient belongs to, from NLM RxClass, deduplicated and labeled.

Historical crypto price, market cap and volume time series. Given a coin id or ticker and a window of days (1-365) returns aligned price, market-cap and total-volume points with ISO timestamps — finer-grained than OHLC candles. For charting, market-cap trend analysis, volume studies, backtesting and portfolio agents.

Crypto derivatives, perpetual futures and funding rates. Returns active derivative contracts with market/exchange, symbol, index, price, 24h change %, funding rate, open interest and 24h volume. For funding-rate arbitrage, perp market monitoring, basis-trading and derivatives-analytics agents.

Crypto market sectors ranked by market cap. Returns categories (e.g. DeFi, Layer 2, meme coins, AI, real-world assets, stablecoins) with total market cap, 24h market-cap change %, 24h volume, top coins and last-updated timestamp. For sector rotation, narrative tracking, market-structure analysis and thematic screening agents.

USD Coin

Get funding rate pattern signals (streaks, extremes, divergences)

Returns current blockchain information including chain height, difficulty, and network statistics.

Prins 2016 PC-PTSD-5: caller-assigned lifetime trauma-exposure gate plus five past-month yes/no items (nightmare/intrusion, avoidance, hypervigilance/startle, numbing/detachment, guilt/blame). Score 0 if trauma_exposure is false (items required but not counted). If true, each yes is 1 (max 5). Sensitive screen at ≥3. Not a PTSD diagnosis or medical device. magent does not interview.

Cox 1987 EPDS: 10 caller-assigned integers 0-3 (past 7 days). laugh, enjoyment, and anxious are reverse-scored in that paper; the caller assigns the published 0-3 (0 = least symptomatic pole after that reverse key). Total 0-30. This paper's threshold is 12/13 (at or above iff score ≥13). magent does not interview, apply later ≥10 community cuts, PHQ-9, QIDS, or MDI. Not a diagnosis.

Chapman 2016 RibScore from six caller-assigned radiographic flags (≥6 rib fractures, bilateral, flail chest, ≥3 severely bicortical displaced, first rib, and all three anatomic areas). 1 point each, total 0-6. Associated linearly with pneumonia, respiratory failure, and tracheostomy. magent does not read a CT. Deterministic published arithmetic with citation; not a named risk class.

Hamilton 1960 17-variable depression rating (not the later 21-item form). Eight caller-assigned 0-4 items and nine 0-2 items (agitation is three-point in that paper). Sum 0-50. magent does not interview. Hamilton 1960 printed no total-score severity bands; later 0-7/8-16/17-23/>=24 cutoffs are not applied. Not a diagnosis and not a medical device.

Hirschfeld 2000 Mood Disorder Questionnaire (MDQ): thirteen caller-assigned yes/no lifetime manic/hypomanic items, 1 point each, max 13. Positive screen requires score ≥7 AND same_period AND moderate_or_serious_problem (that paper: sensitivity 0.73 and specificity 0.90 in a psychiatric outpatient sample). Not a bipolar diagnosis, not a medical device, and not Hirschfeld 2003 general-population cutoff tweaks. magent does not interview the patient.

Hamilton 1959 HAM-A: 14 caller-assigned integers 0-4 (0 not present, 1 mild, 2 moderate, 3 severe, 4 very severe) for anxious mood, tension, fears, insomnia, intellectual, depressed mood, somatic muscular, somatic sensory, cardiovascular, respiratory, gastrointestinal, genitourinary, autonomic, and behaviour at interview. Sum 0-56. magent does not interview the patient. No total-score severity bands in Hamilton 1959; later 17/24 cutoffs are not applied. Not a diagnosis and not a medical device.

Bellera 2012 G8 geriatric screening tool: seven MNA items (A, B, C, E, F, H, P) plus age. Food intake 0–2, weight loss 0–3, mobility 0–2, neuropsychological 0–2, BMI 0–3, more than 3 medications 0 if true else 1, health vs peers 0/0.5/1/2, age 0–2. Total 0–17, higher is better. Score ≤14 is an impaired screen. Not a CGA, chemotherapy order, Soubeyran 2014 ONCODAGE, or VES-13. magent does not interview or weigh the patient.