NeverEmpty

For property investment screening, market reports and agency lead lists: Rightmove.co.uk listings as flat rows with price and what it means, bedrooms, type, address, coordinates and agency. London alone had 59,361 for sale and 27,624 to rent on 2026-09-05. Monitoring returns only new listings.

Give it feed URLs or plain website URLs. It finds the feed a site advertises, reads RSS 2.0, RSS 1.0/RDF, Atom and JSON Feed in any encoding, and returns one row per item with title, link, dates, author, summary, full text, categories, image and enclosures. Monitoring mode returns only new items.

For KYC, AML onboarding and audit files: screen names against the official OFAC SDN, EU and UN lists, one row per name with the alias it matched, how close, and which list version. 19,319 OFAC, 6,234 EU and 1,011 UN records, 78,733 names with aliases on 2026-08-27. A clean name is an answer.

Every Form D private placement filed with the SEC EDGAR, one row per filing: issuer, industry, amounts offered and sold, investors, Rule 506(b)/(c), first sale date and officers. Filter by date, state, industry and size. Indefinite amounts stay null, never 0. No API key.

For SEO agencies who need findings as data not a PDF, developers wiring a check into CI, and AI agents: one row per page with title, meta description, canonical, robots headers, headings, image alt, JSON-LD and hreflang. No invented score - every finding names the tag it came from.

For SEO audits, crawl seeding and keeping a RAG index fresh: give it a domain and get one row per URL with lastmod, change frequency and priority. Asking for /sitemap.xml fails on real sites - stripe.com answers 404 - so it reads robots.txt, follows nested indexes and unpacks gzip. No API key.

For recruiters, talent intelligence and competitor hiring tracking: every public job of any company on SmartRecruiters - Bosch, Ubisoft, CERN and thousands more - one row per job with title, department, city, country, remote and hybrid flags and level. A wrong ID is never reported as 0 jobs.

For betting models, fantasy tools and sports dashboards: live, finished and upcoming games for MLB, NHL, NFL, NBA, NCAAF, Premier League, La Liga and MLS in one identical row shape, from official league APIs. Status folds into pre, in and post. A league out of season gets a no-games row, not silence

For OSINT, crypto and market research and channel vetting: posts, views, reactions and forwards from any public Telegram channel, plus the exact subscriber count from the profile page. On one 10.9M channel the rounded preview hid 43,313 people. Median views, ERR and cadence are measured.

For betting models, backtesting and tennis analytics: every TennisExplorer match as a flat row with set-by-set scores, tiebreaks, decimal odds on both sides and seeds. A busy day is 224 matches and 212 of them carry odds; any past day back to 2016 still returns. Export as JSON, CSV or Excel.

For MMA betting models, fantasy tools and analytics: official ufcstats.com data covering 788 events back to UFC 2 in March 1994. Every fight round by round, with significant strikes split by head, body and leg and by distance, clinch and ground, takedowns, control time and judges' scorecards.

For bid teams and public-sector sales: UK tenders and contract awards from both official services at once, Contracts Finder and Find a Tender. On 2026-09-09 alone that was 580 notices (136 plus 444). Buyer, CPV, value, deadline, region and winning supplier, duplicates merged. No API key.

For airline ops, travel apps and freight scheduling: every US airport the FAA reports a problem at - ground stops, ground delay programs, closures, arrival and departure delays - with delay lengths already in minutes. Its feed repeats sections: taking one block returned 3 of 5 closed airports.

For contractor lead gen, construction suppliers and property analysts: newly issued building permits from 8 official city open-data portals in one schema. New York alone issued 1,829 permits in a 7-day window. One row per permit, and value filters compare numbers, not text. No API key.

For trading signals, journalism and ethics research: every stock trade US House members disclose under the STOCK Act, from the Clerk of the House - ticker, buy or sell, trade and notification dates, whose account, and the disclosed range as lower and upper numbers. 2026 holds 1,603 filings.

For importers, customs brokers and landed-cost tools: official USITC tariff lines with duty rates as numbers you can calculate with. 10-digit statistical lines come back with a rate - in the raw schedule all 35 lines under heading 6109 are blank. No API key.

For flood risk models, water utilities, insurance and paddling apps: river level, streamflow and water temperature from official USGS gauges, with how old each reading is and whether the sensor worked. Over 31,752 time series on 2026-08-30, 2.64% read -999999 and 26.6% were over a year old.

For site migrations, redirect maps, recovering deleted pages and domain diligence: every snapshot of a URL, or every archived URL under a domain, from the official Wayback CDX API - first and last capture, capture count, status codes and a link to the copy. One page alone can return 768,693 rows.

For lead generation, outbound sales and agency prospecting: emails, phone numbers and social profiles actually published on a list of websites, one row per site in the order you gave them. It will not invent an address - 'the team at dealer.com' is not an email, nor a VAT number a phone number.

For property investment screening, agency lead lists and market reports: willhaben.at listings as flat rows with price and what it means, price per m2, rooms, area, postcode and coordinates. Vienna alone had 20,089 flats for sale on 2026-09-05. Monitoring returns only listings new to you.

For RAG pipelines, LLM training sets and content research: give it @handle, a channel URL or a UC id and get transcripts for that channel's recent videos - no collecting video IDs with a second tool first. 180 videos across 12 channels measured on 2026-08-29. Export as JSON, CSV or Excel.

For channel tracking, competitor research and AI datasets: give it a handle and get that channel's videos with views, length and publish time - read only through paths YouTube's robots.txt allows. The RSS feed most scrapers use is disallowed; the allowed page gave 30 videos where RSS gives 15.

For sentiment analysis, community management, creator research and AI training sets: comments and replies from any public video with text, author, channel, likes, replies and age. YouTube rounds like counts before sending them - 42 of 68 measured on 2026-09-08 - so every row says if it is exact.

For influencer vetting, sponsorship reporting and trend research: Shorts from any YouTube channel or search term with id, URL, title, channel, thumbnail and the exact view count. The number printed under a Short is truncated, not rounded - one shown as 10M really had 10,962,867 on 2026-09-08.