Brains

Knowledge somebody else already did the reading for.

Public brains built by the Ounie community. Ask one a question and the answer comes back cited to its own sources — fork the free ones, buy into the paid ones, or point your agent at any of them.

  • Every answer cites its sources
  • Fork the wiki, not the files
  • Agents read and pay over MCP

Showing 23 of 23 brains

One dated edition per trading day: SEC Form 4 cluster buys — two or more insiders independently buying the same issuer on the open market — ranked by a deterministic 0–100 conviction score with the arithmetic shown.

5 cr / sourcex402
@ounie11 sources2

One dated edition per city per week: building-permit velocity by neighborhood, scored against that neighborhood's own trailing year. Every surge has to be unusual twice over — an unusual share of the city AND an unusual week for the area — so a citywide holiday cannot manufacture one. Austin, Chicago and New York, from the cities' own open-data portals, with the arithmetic shown.

5 cr / sourcex402
@ounie8 sources

One dated edition per day: every CVE published to NVD, plus that day's CISA KEV additions, triaged into patch today / this week / watch by a deterministic 0–100 score built from known exploitation, EPSS probability, CVSS severity, network exposure and blast radius — grouped by stack, with the arithmetic shown.

3 cr / sourcex402
@ounie14 sources

One dated edition per state per day: every business entity registered yesterday in Colorado and New York, cut into the segments a salesperson actually works — owner-run construction LLCs, new health practices, and so on. Foreign qualifications and same-day dissolutions are removed, and mail drops are identified by how many of the day's other filings share the address, so a wall of virtual mailboxes never leads the page.

5 cr / sourcex402
@ounie9 sources

One dated edition a week on what was filed at the Federal Energy Regulatory Commission: transmission and pipeline rate changes, gas certificate applications, hydro licensing, complaints and Commission orders — cut into the segments an energy analyst or a ratepayer advocate actually works. About half of every week is routine quarterly paperwork; it is counted and kept out of the listings rather than padding them. FERC is wholesale and interstate: retail rate cases stay at the state commissions and are not covered.

10 cr / sourcex402
@ounie4 sources

Every Federal Reserve Board speech, testimony and FOMC statement, scored on a fixed hawk/dove lexicon so the numbers mean the same thing in every edition, plus the sentence-level diff between consecutive FOMC statements. Each speaker's stance is compared with their own prior month, because the movement is the signal. More than half of Fed speeches are about payments, AI or regulation rather than policy; those are listed and deliberately left unscored rather than given a meaningless lean.

5 cr / sourcex402
@ounie5 sources

One dated edition a week of what actually happened on ClinicalTrials.gov: trials newly registered, trials halted, trials completed, and the FDA approvals that landed alongside them. Every halt carries the sponsor's own stated reason, classified — because a registry status cannot tell a drug that failed from one that worked so well the trial stopped early, and it cannot tell either from a study paused for the summer. Reasons the sponsor did not give are marked as not given, never guessed.

10 cr / sourcex402
@ounie3 sources

One dated edition a week on what members of Congress disclosed trading, from the House clerk's and the Senate's own STOCK Act filings. Tickers two or more members traded the same way are grouped and scored; every amount is reported as the band the law actually requires rather than a made-up midpoint. Includes the disclosure lag on every row — some run past a year — and names the filings that arrived as scanned paper and could not be read.

5 cr / sourcex402
@ounie4 sources

One dated edition per filing day: every US private funding round reported to the SEC on Form D, with the pooled investment funds stripped out, segmented by the issuer's own declared industry, and each round shown with the amount sold, the date of first sale and a link to the filing. Amendments, announced-but-unsold offerings and late filings are reported separately rather than ranked, because a Form D amendment's total is cumulative over the offering's whole life and would otherwise read as a raise.

10 cr / sourcex402
@ounie6 sources

Everything about Ounie — what it is, how it works, and how to use it — answered from the source, with citations. Covers the second-brain concept, the LLM-Wiki + RAG architecture, adding sources, the wiki and knowledge graph, sharing and forking, plans and pricing, credits, widget

Free
@ounie13 sources2

One dated edition per month: every industry-sponsored Phase 2 and Phase 3 trial whose primary completion falls in the next two months, grouped by phase and graded for how much the date is worth believing. Every forward-looking completion date in the registry is the sponsor's estimate, so the grade — built from whether enrolment has closed, how far the trial has got, whether the sponsor named a day or only a month, and when the record was last revised — is the product. Withdrawn and suspended trials are kept off the calendar and listed separately. No PDUFA dates, no tickers, no probability of success: the registry publishes none of those.

10 cr / sourcex402
@ounie2 sources

One dated edition a week per city on what the health department actually did: every closure and failed inspection as a record, with the city's own enforcement rate scored against its trailing year so a sweep is not mistaken for a wave. Names are grouped by footprint — a citation against a two-hundred-address brand reads differently from one against a single shop — and repeat closures are counted from the department's own record. There is no restaurant score and no restaurant ranking anywhere in this feed, deliberately: an inspection record is a fact, not a verdict.

3 cr / sourcex402
@ounie4 sources

One dated edition a week of every federal funding opportunity newly posted on grants.gov, cut by who is actually eligible rather than ranked — a nonprofit gets the list a nonprofit can apply to. Award sizes are printed as the range the agency published, never a made-up midpoint, and the quarter that publish no figure are marked unspecified. Agency forecasts, which have no deadline yet, are kept in their own section instead of being given a fake one. Deadlines inside 30 days lead the edition.

10 cr / sourcex402
@ounie2 sources

One dated edition a week per ecosystem on what actually moved: stable releases shipped, commit volume against each project's own trailing year, how many distinct people committed, issues closed against issues opened, and the projects drifting toward abandonment. Star counts are printed as context and are not an input to any verdict — release cadence, contributor breadth and issue throughput are what the rankings rest on. Each ecosystem is a fixed list, so two editions are comparable.

5 cr / sourcex402
@ounie6 sources

One dated edition per state per month: the ZIP codes where HUD's Small Area Fair Market Rent pays MORE than the open market, so a landlord's rent rises rather than falls by accepting a Housing Choice Voucher. Built from HUD's published SAFMR schedule against Zillow's rent index, anchored at the bedroom count the two sources actually agree on, with the dollar spread, the annual figure and the arithmetic shown for every ZIP.

25 cr / sourcex402
@ounie3 sources

One dated edition a week on where North American freight rail got slower, from the Surface Transportation Board's own EP 724 filings. Train speed, terminal dwell, cars on line, trains held and cars stalled 48 hours or more, each scored against its own trailing year AND against its peers — so a blizzard that slows every railroad is reported as one storm rather than six failures. Includes the worst terminals in absolute hours, which a movers table cannot show, and names every series that did not report.

10 cr / sourcex402
@ounie3 sources

One dated edition a week per jurisdiction on where housing pressure moved, from court and marshal records published as open data. Every neighbourhood is measured twice — against its own trailing year and against its share of the jurisdiction — so a statewide seasonal swing is never printed as sixty neighbourhoods in crisis. New York's executed warrants and Connecticut's filings are reported separately and never added: one is the end of a process and the other its beginning. No person appears anywhere in this feed; the unit is an area, always.

10 cr / sourcex402
@ounie4 sources

One dated edition a month on the Single Audits just accepted by the federal Audit Clearinghouse: which cities, counties, school districts, tribal governments, universities and federally-funded non-profits filed an audit raising going-concern doubt or material noncompliance, named, with the federal money at stake and how late the audit was. Segmented by state, entity type and spend band, with each finding rate scored against its own trailing year. Archive-dominant: an organisation's run of audits is the product.

15 cr / sourcex402
@ounie3 sources

One dated edition a week per App Store chart and genre on what actually broke out: rank velocity week over week, with breakouts defined as two consecutive climbing weeks rather than one promoted Tuesday. New entrants, hard falls and exits are reported separately, an exit is never given an invented magnitude, and every edition prints the genre's own churn so three movers in a volatile chart are not read like three in a still one. Apple publishes only the present, so this archive builds forward.

Free
@ounie1 source

Persuasion frameworks and copy principles from marketing legends — Hopkins, Ogilvy, Bernbach, Hormozi, Godin, and more. The structure layer for the Landing Page Builder: how to say it, not what's true. Pulled per-page to shape headlines, value, proof, objections, and CTAs.

ForkableFree
@ounie13 sources13

Publish one of your own

Build a brain from your sources, make it public, and let anyone — or any agent — ask it. Price it per question or by the source; every answer cites your material back to you.