Search individual US Congress member stock trades, parsed from STOCK Act Periodic Transaction Reports into clean rows. Filter by member name (q, e.g. "Pelosi"), ticker (e.g. NVDA), transaction type (purchase | sale | exchange | partial_sale), party (D | R | I), 2-letter state, chamber, or transaction-date range (dateFrom/dateTo). Each trade returns the member + party + state/district, owner (self/spouse/joint/dependent), ticker + asset description, buy/sell, the disclosed dollar RANGE (amountMin/amountMax — disclosures are ranges, not exact), the transaction date, the disclosure date, days-to-disclose (compliance lag), and a link to the source filing PDF. The envelope total is the full count matching your filter — so ticker=NVDA&type=purchase counts who bought NVIDIA. Amounts are ranges by law; trades are disclosed up to 45 days after they happen (current-to-the-filing). Coverage: US House + Senate e-filed PTRs; scanned/handwritten reports are parsed separately. Data: US House Clerk / US Senate eFD (public domain). Cross-ref /api/gov/congress-filings for the raw filing index.

Live perp funding rates, open interest, and mark/oracle/mid prices across 200+ Hyperliquid perpetuals (free/keyless). Each row: coin, hourly funding rate, open interest, mark/oracle/mid price, premium, prior-day price, 24h notional volume, max leverage. Filter by coin; sort by oi, volume, or funding. On-chain perp microstructure for funding-arb and OI signals.

A Polymarket trader's portfolio by wallet address: total portfolio USD value plus open positions (market title, outcome, size, average vs current price, current value, and unrealized PnL). For tracking a wallet's prediction-market book. Read-only from Polymarket's Data API.

Compute the current position of any cataloged Earth-orbiting satellite by NORAD catalog number (e.g. 25544 = ISS, 20580 = Hubble), using fresh Celestrak orbital elements propagated with SGP4. Returns sub-point latitude/longitude, altitude (km), and speed (km/s) for the requested instant (default: now; pass `at` as an ISO timestamp for any time within a few days of the elements' epoch). Supply observer `lat`/`lon` (and optional altKm) to also get look angles — azimuth, elevation, slant range, and whether the satellite is above your horizon — for visual/antenna pointing. Satellite tracking, pass prediction, ground-station planning. Public-domain orbital data; physical/mission metadata is not included.

Cross-venue equivalent-market pairs for arbitrage / relative-value spotting. We fuzzy-match open Polymarket vs Kalshi markets by title-token overlap and return candidate pairs with both venues' YES prices and the implied-probability spread (Polymarket YES minus Kalshi YES). Tune with minScore (0-1 title-similarity floor) and limit. Heuristic — a high score is a strong candidate, not a guaranteed identical contract. Read-only, keyless. No public source maps these directly; the pairing is derived by 2s.io.

Resolve any organism by scientific or common name to the GBIF taxonomic backbone — the canonical free reference for life on Earth. Returns the accepted scientific name + canonical name, taxonomic rank and status, the match type/confidence, the full classification (kingdom → phylum → class → order → family → genus → species), up to 8 English vernacular (common) names, the count of recorded occurrences worldwide, and the GBIF species URL. Biodiversity, ecology, education, and as a taxonomy normalizer for other pipelines. Fuzzy-matches misspellings. CC-BY GBIF data.

Parse, validate and canonicalize a phone number using Google's libphonenumber metadata. Query: number (required; E.164 like "+14155552671" or national like "(415) 555-2671"), country (optional ISO 3166-1 alpha-2 like US/GB/JP — required when number is not E.164). Returns {input, valid (matches a real number range), possible (right shape/length), e164, national, international, rfc3966, country, countryCallingCode, type (mobile|fixed_line|fixed_line_or_mobile|toll_free|premium_rate|shared_cost|voip|personal_number|pager|uan|voicemail|unknown), possibleCountries[]}. Returns valid:false (not an error) for unparseable input.

Find the right 2s endpoint for a task using natural language. Pass q (e.g. 'check if a domain can be spoofed', 'is this CVE being exploited', 'screen a company for sanctions', 'decode a VIN') and get back the catalog's most relevant endpoints, ranked, each with its id, API path, method, group, description, and price. Deterministic keyword + synonym matching over the live endpoint registry (no LLM, no external calls) — a self-routing index so an agent (or its planner) can discover which paid call answers a question instead of reading the whole 290+ endpoint catalog. Returns the path you then call directly.

Historical quarterly earnings surprises for a US-listed company — reported (actual) EPS vs the analyst consensus estimate, the absolute surprise, and the surprise percentage, for the most recent quarters (newest first). Pass ticker (optionally limit). Tells you whether a company has been beating or missing expectations. Data by Finnhub.

UK street-level crime around a location from the Home Office data.police.uk service. Give a latitude/longitude (and optional month YYYY-MM, defaults to the latest available — UK data lags ~2 months) and get the total crime count, a breakdown by category (anti-social behaviour, burglary, violent crime, vehicle crime, etc.), and recent individual records (category, approximate street, lat/lon, and outcome status). Free, Open Government Licence (commercial use permitted). Location-risk signal an LLM cannot recall — for property, insurance, and safety agents. Locations are anonymised to nearby map points by the Home Office.

One bounded batch uses the existing technology detector with a fresh fetch per domain. Reports additions, removals and version changes only when a caller supplies a previous snapshot. First scans have changes=null. Root HTTPS URLs only. Caller snapshots are not authenticated historical evidence. No background jobs, shared history, callbacks or inferred installation dates. Fixed price for a batch of one to three domains.

The US economic-data release CALENDAR — when official economic reports are published, from the Federal Reserve's FRED. Returns upcoming release dates (date, release name, FRED release id), starting from `from` (default today), ascending. Filter by name to find a specific report's schedule (e.g. name="Consumer Price Index" → the next CPI dates; "Employment Situation" → jobs report; "Gross Domestic Product"; "FOMC"), or by releaseId. Free, public-domain (FRED). Future release dates an LLM cannot possibly know — essential for trading, scheduling, and macro-monitoring agents that need to act around data drops. Pair with econ.fred to pull the actual numbers once a release is out.

Look up or list US Congressional hearings via Congress.gov. Pass congress + chamber + jacketNumber to fetch a single hearing with full detail (title, dates, committee, transcripts/recordings if available). Or filter list by congress + chamber, with optional date range. Hearings cover committee testimony from the executive branch, subject experts, and stakeholders — primary-source material for oversight + policy research.

Look up an ISO 4217 currency. Pass code (alphabetic like USD, or 3-digit numeric like 840) for the canonical record, or country to find the currency/currencies a country uses. Returns the alphabetic + numeric code, English name, minor unit (decimal places — e.g. JPY 0, USD 2, BHD 3), and the countries using it. Bundled authoritative ISO 4217 data — the currency-code + decimal-places ground truth an LLM gets wrong on less-common currencies.

Scrape Creators · Truth Social — User Posts

Current weather conditions for any coordinate worldwide. Pass lat + lon (and optional units=metric|imperial, default metric). Returns temperature, feels-like, humidity, precipitation (rain/snow), cloud cover, sea-level pressure, wind speed/direction/gusts, day/night flag, and a human-readable conditions label from the WMO weather code — with local time + timezone. Global coverage (not just US — for US-only NWS data see weather.zip / weather.forecast). Source: Open-Meteo (CC BY 4.0).

Terminal Aerodrome Forecast (TAF) for one or more airports — the official aviation forecast issued for an airport (~24–30h ahead). Pass ids = comma-separated ICAO station ids (e.g. KATL). Returns the raw TAF text + issue time. Source: NOAA Aviation Weather Center (keyless). For current conditions use aviation.metar.

Aggregate financial summaries (receipts, disbursements, cash-on-hand, debt, transfers, refunds, contributions by source, etc.) per cycle for federal candidates (scope=candidates) or committees (scope=committees). Filter by candidateId / committeeId, cycle, office (P/S/H), party, state, district. For candidates, electionFull=true rolls all cycles of a single election into one row (e.g. a presidential cycle spans 4 years). Use this for top-line "how much did they raise/spend" answers without paging through millions of itemized transactions.

Search OSHA inspection records via the US Department of Labor Open Data Portal (~5M historical inspections). Filter by US state (2-letter), city, zip, or establishment-name substring (estabName, case-insensitive). Optional `filter` accepts raw OData-style clauses (joined with AND); `fields` selects columns; `sort` sorts (prefix with `-` for desc). Each row carries activity number, reporting id, establishment name + full site address, owner type/code, advance-notice flag, safety/health type, SIC + NAICS industry codes, open + close dates, host establishment number, union status. Backbone of workplace-safety + investigative-journalism agents.

USDA FoodData Central nutrition lookup. Two modes: search (query=cheddar cheese) returns matching foods with fdcId, description, data type, category, and brand; detail (fdcId=328637) returns the full analyzed nutrient profile — energy, protein, fats, carbohydrates, vitamins, minerals — with USDA nutrient numbers, amounts, and units (per 100 g/ml unless serving fields indicate otherwise), plus ingredients for branded foods. Filter search by dataType: Foundation and SR Legacy are lab-analyzed generic foods, Branded is the packaged-food label corpus, Survey (FNDDS) is as-consumed dishes. Authoritative USDA data covering ~400k foods — use real analyzed values instead of model-estimated nutrition facts.