KeyVex

Returns Schedule 13D / 13G beneficial-ownership disclosures — filings made by anyone holding ≥5% of a class of registered equity securities. Each record is one reporting person on one filing (joint filings emit multiple rows under the same accession_number). Use this when the user asks about: who's accumulating large stakes, activist campaigns, takeover targets, hostile bids, or institutional concentration in a name. Also for 'who owns this company at the 5%+ level?' questions. Two flavors, distinguished by `is_activist`: - **13D (is_activist=true)**: filer signals INTENT TO INFLUENCE control. Activist campaigns, takeover stakes, hostile bidders. - **13G (is_activist=false)**: filer is PASSIVE. Mutual funds, advisers, banks, insurers, qualified institutional holders. Filter `is_activist: true` to see only the takeover-style filings — much higher signal-to-noise than the 13G firehose, which is dominated by routine quarterly disclosures from Vanguard, BlackRock, etc. COVERAGE FLOOR: KeyVex's ingestion of Schedule 13D/G filings begins January 2024. Filings before 2024 are not in the collection. A 13D/G query for activity in 2023 or earlier returns zero records — this is the collection's coverage boundary, not a per-entity gap. 13D/A and 13G/A AMENDMENT EXIT FILINGS: when a filer reports they have divested below the 5% threshold, the resulting row carries shares_owned: 0 and percent_of_class: 0. These rows are CORRECT — an exit IS zero — not missing data. To distinguish active stakes from exit filings, filter by shares_owned > 0 or by min_percent_of_class. ⚠ THIS TEXT USED TO SAY Item 4 WAS NOT AVAILABLE HERE. It is, as of 2026-08-28. Pass include_filing_narrative:true for the filer's own 'Item 4: Purpose of Transaction' — where board seats, consent rights and control intentions are actually declared — plus Item 2 (the ownership chain behind the filer), Item 2(d)-(e) (criminal and civil proceedings, as prose), Item 3 (source of funds) and Item 6 (the contracts with the issuer that implement Item 4). It is off by default only because it is long, not because it is missing: measured at 58.5% of a row. Rows holding withheld narrative say so via filing_narrative_available. 13D only.

Returns FAA aircraft registrations — the releasable registry of ~314K currently US-registered aircraft, one record per N-number: the aircraft (manufacturer, model, seats, engines, weight class, year built, engine type), and the REGISTRANT — name, type (Individual / Corporation / LLC / Government / Co-Owned, the FAA's own legend, with the verbatim code), address, and up to five co-owner names. Use this when the user asks: who owns an aircraft (by N-number), what aircraft a company / person / nonprofit registers, corporate-jet fleets, aircraft registered in a state, or to join tail numbers seen in flight-tracking data to owners. Lookups: n_number is a direct hit ('N123AB' or '123AB'). registrant_name matches the primary registrant AND co-owner names. Note: many corporate jets register through trustee banks (e.g. 'BANK OF UTAH TRUSTEE') or aircraft-management LLCs — an absent company name is NOT proof the company has no aircraft. Registry posture: a CURRENT-roster snapshot refreshed monthly from the FAA's daily-updated file. snapshot_date is the last snapshot a record appeared in — stale means since deregistered. Deregistered- aircraft history (the FAA DEREG file) is a planned follow-up. Pure-publisher posture: FAA records as published, parsed by KeyVex; type labels are the FAA's own documentation legends, never our inference.

Your watchlist alerts, newest first: events matched to the tickers, members of Congress and federal-contract recipients (UEIs) on YOUR watchlist, for the API key making the call. Kinds: congress_trade: a congressional trade disclosed this week in a ticker or by a member on your watchlist. initial_13d: an initial Schedule 13D (a new 5%+ activist stake) filed this week in a ticker on your watchlist. federal_contract: federal awards at or above your threshold that KeyVex first reported this week, new or newly modified (a later modification of an award you were already alerted to does not alert again). An event alerts only when its own date (disclosure, filing, the award's latest modification) is within 7 days of detection and not before you added the entry, so a watchlist never replays history. Each alert's `data` is the source record as KeyVex stored it when the alert fired, with KeyVex's internal `_` fields removed; the matching tool may present the same record differently. Paginate with next_cursor. The watchlist itself is managed on keyvex.com. Webhooks (paid plans, Basic and up): each POST carries X-KeyVex-Timestamp and X-KeyVex-Signature = sha256 HMAC of '<timestamp>.<raw body>' with your secret. Receivers MUST verify the signature and REJECT a timestamp older than 5 minutes (replay protection); redirects are never followed; answer 2xx within 5 s.

Returns Form 278 (Public Financial Disclosure / Annual Financial Disclosure) filings — the annual snapshot members of Congress file each year showing assets, income sources, liabilities, transactions, gifts, outside positions, and (for spouse + dependent children) the same. The same filings are published free as news at https://keyvex.com/disclosures under 5 U.S.C. § 13107(c). SCOPE — v1 covers BOTH chambers: Senate (Senate eFD) and House (House Clerk). Filed by every senator and representative (and senior executive-branch officials, federal judges) by May 15 each year. Use this when the user asks about: a member's asset composition, outside income sources, board seats / outside positions, liabilities (mortgages, loans), or for news reporting on annual disclosures. CONTENT — when a filing's schedules were machine-parsed, `content_parsed` is true and the record carries structured `assets` (Schedule A) and `liabilities` arrays plus `asset_count` / `liability_count`. `value_range` / `amount_range` are the disclosed RANGES (e.g., '$50,001 - $100,000'), NOT point estimates — KeyVex does not collapse a range to a single number. SENATE rows carry the ranges verbatim. HOUSE rows are read from the House Clerk PDF by column position. On a House candidate or new-filer report, whose income prints in two columns ('current year to filing', 'preceding year'), `income_range` is empty — neither is the reporting period; `report_url` shows both. A row KeyVex could not read with confidence carries `parse_unreliable: true`, and for such rows `report_url` is authoritative. Net-worth roll-up is intentionally NOT provided (it would be a KeyVex-derived aggregate, not a disclosed value). When schedules are unavailable — Senate PAPER (scanned-image) filings, which carry no machine-readable text, or the occasional parse skip — `content_parsed` is false and `coverage_note` names the limitation; follow `report_url` to read the original. This honest coverage boundary is never a silent omission. Different from get_congressional_trades: PTRs are per-trade real-time notices (filed within 30-45 days), while Form 278 is the year-end balance-sheet snapshot. Combine both for the full activity + position view of a member. Report types: 'Annual' (yearly filing covering prior calendar year), 'New Filer' (initial disclosure on entering office), 'Termination' (final disclosure on leaving office), 'Combined' (annual+termination for filer who left mid-year), 'Amendment' (correction of a prior filing), 'Other' (rare).

Returns quarterly financials for every FDIC-insured US bank — FDIC BankFind Suite (RIS) data derived from Call Reports, one record per (bank, quarter-end) from 1984 to the present. Covers ~4,400 active institutions per modern quarter, most of which never file with the SEC — this is the banking-sector complement to get_fundamentals. Use this when the user asks about: a specific bank's assets / deposits / profitability / capital ratios, bank league tables ('largest banks in Texas'), deposit flight or brokered-deposit reliance, nonperforming-asset trends, or to pair bank fundamentals with OCC/FDIC/Fed enforcement actions and CFPB complaints. Record shape: identity (cert = FDIC certificate number, the stable bank key; name/city/state), balance sheet in $ THOUSANDS (total_assets, total_deposits, equity_capital, net_loans_leases, securities, loan buckets, brokered_deposits, insured-deposit estimates), income statement in $ thousands (net_income, interest income/expense, noninterest income/expense, provisions), and FDIC-computed ratios in percent (roa, roe, net_interest_margin, efficiency_ratio, leverage_ratio, tier1_risk_based_ratio, cet1_ratio, total_risk_based_ratio, nonperforming_assets_ratio, net_chargeoffs_ratio, loans_to_core_deposits). CONVENTIONS (Call Report): income-statement items are YEAR-TO-DATE (Q3 net_income = nine months, not the quarter alone; Q4 = full year); ratios are FDIC's annualized computations; roe_quarterly is the single-quarter ROE. Balance-sheet items are point-in-time. Matching a bank: cert (FDIC certificate #) is exact and stable — resolve it once via a name substring query, then use cert for history. Banks are subsidiaries: 'JPMorgan Chase Bank, National Association' (cert 628) is the insured bank, not the NYSE-listed holding company — for the parent's SEC financials use get_fundamentals. League tables: report_date='<quarter-end>' + sort_by='total_assets' (e.g., report_date '2026-03-31'). Quarter-ends are 03-31 / 06-30 / 09-30 / 12-31. Latest quarter fills in progressively for ~60 days after quarter end as banks file. Pure-publisher: FDIC's numbers as published, no derived health scores.

Returns congressional bill metadata from api.congress.gov. Use this when the user asks about: bills introduced this Congress, the status of a specific bill, House vs Senate bill volume, what bills mention a topic, or to bridge from a roll-call vote (legislation_type + legislation_number) to the underlying bill. Source: api.congress.gov v3 (Library of Congress). Covers ALL bill types: HR (House Bill), S (Senate Bill), HRES (House Simple Resolution), SRES (Senate Simple Resolution), HJRES (House Joint Resolution), SJRES (Senate Joint Resolution), HCONRES (House Concurrent Resolution), SCONRES (Senate Concurrent Resolution). v1A returns metadata only: title, type + number, originating chamber, latest action (date + text), and links. Sponsors, cosponsors, full action history, bill text, and CRS summaries live at `api_url` (structured JSON) and `congress_gov_url` (public HTML). Agents follow those for prose detail. Bill identifiers are stable composite keys formatted as {congress}-{TYPE}-{number}, e.g., '119-HR-134', '119-S-1234', '118-HJRES-5'. Use bill_id for the fastest direct lookup. Pure-publisher posture: KeyVex returns what's in the public record. No legislative outcome predictions, no 'likely to pass' signals.

Returns CFTC Commitments of Traders (COT) report rows — weekly aggregated futures + options-on-futures positioning by trader class. The COT report is the macro positioning dataset for U.S. futures markets. Released every Friday 3:30 PM ET for the prior Tuesday close. Trader classes (the CFTC's Legacy report, futures and options combined): - Non-commercial (large speculators — hedge funds, CTAs) - Commercial (hedgers — producers, swap dealers) - Non-reportable (small speculators) Every row carries report='legacy' and variant='combined' (futures and options together). Killer query patterns: - Macro positioning snapshot this week: latest_only=true (gives the latest report row for every contract in one query) - Large-spec extremes in S&P: commodity_name='S&P 500 STOCK INDEX' + sort_by='noncomm_net' + sort_order='desc' - Gold positioning history: commodity_name='GOLD' + since='2026-01-01' - Currency COT: contract_market_name substring 'YEN' / 'EURO' Source: publicreporting.cftc.gov/resource/jun7-fc8e.json, the CFTC's Legacy Futures-and-Options Combined report (Socrata API, free, unauthenticated); each row's source_url is that row's own query on it. Covers EVERY regulated U.S. futures + options- on-futures contract — agricultural commodities, metals, energy, financials, FX, crypto. Pure-publisher posture: raw positioning numbers, no derived sentiment scores. Key derived fields (computed from raw): noncomm_net (large-spec net), comm_net (hedger net), nonrept_net (small-spec net). Concentration fields show top-4 / top-8 trader net long/short concentration.

Returns ONE company's reference profile: legal name, CIK, tickers + exchanges, SIC industry classification, state of incorporation, HQ and mailing addresses, phone, fiscal year end, EIN, former names, a plain-English business summary (condensed from the latest 10-K Item 1), CEO + board of directors (from the latest DEF 14A proxy), federal- contract activity summary (USAspending join, with a ready-to-run get_federal_contracts query), company logo reference, and — on paid plans — the latest end-of-day closing price (price_eod; Close Prices from Tiingo.com). For price HISTORY use get_daily_prices. The price_iex_last / price_tngolast intraday fields remain unlicensed stubs. ⚠ ASSEMBLED REFERENCE DATA — unlike KeyVex's mirror datasets, this profile is NOT byte-faithful government mirroring. Fields are assembled from multiple bases and each carries its own provenance envelope: source, source_type (mirrored | derived | self_collected | asserted | licensed), fetched_at, as_of, and per-field meta (e.g. the SEC accession number every derived field traces to). A field that cannot be established is UNKNOWN with a machine-readable reason; a field staler than its declared cadence reads as INSUFFICIENT_DATA (value retained for knowing consumption). Factual data only — no ratings, rankings, scores, or opinions. Leadership staleness note: CEO/board come from the latest annual proxy and can lag mid-year changes; meta.leadership_change_events_since_proxy lists any 8-K Item 5.02 (officer/director change) filings made after the proxy — follow up with get_material_events for the detail. Lookup is by ticker OR cik (exactly one required). Ticker matches the company's current EDGAR-listed tickers; renamed/delisted tickers miss — use the CIK. An ambiguous ticker (multiple CIKs) returns an error naming the candidate CIKs rather than guessing.

Returns trade records disclosed by U.S. members of Congress under the STOCK Act — Senate eFD and House Clerk Periodic Transaction Reports (PTRs). Each record is one disclosed transaction by a member or their immediate family. The same filings are published free as news at https://keyvex.com/disclosures under 5 U.S.C. § 13107(c). Use this when the user asks about: who in Congress traded a specific stock, what trades a specific member made, recent congressional trading activity, or filings within a date range. Important: This data is *disclosed* trades, with reporting lag up to 45 days. The disclosure_date is when the public could first see the trade; the transaction_date is when the trade actually happened. For 'what did Congress just disclose buying' questions, sort by disclosure_date. For 'what did Congress hold around a specific market event', filter by transaction_date. Each amount is a range like '$1,001 - $15,000' (Senate filers report ranges, not exact amounts). The amount_min and amount_max fields parse those bounds for filtering. STOCK Act allows reporting in 11 standard ranges from $1,001 up to over $50,000,000. Each record's party is the party the member held on the date of the record (a trade's transaction date, a disclosure's filing date); for a date outside the member's terms in office, the party of their nearest term (the last one before that date, or the first one after it).

Returns consumer complaints filed with the Consumer Financial Protection Bureau (CFPB). Each record is one filing against a bank, credit reporting agency, mortgage servicer, debt collector, fintech, or crypto firm — with company response status, timeliness flag, and (when consented) consumer narrative. Use this when the user asks about: complaint volume against a specific company, top issues at a credit reporting agency, regional complaint patterns, untimely responses by a financial institution, or as a leading indicator of upcoming CFPB/OCC/FDIC enforcement action. COVERAGE — live passthrough (source:'live'): each call queries CFPB's own search API over the FULL 15.7M+ complaint database, full history, current as of CFPB's publication. The response's `total_count` is CFPB's authoritative count for your filtered query — USE IT for volume answers (the `results` array is just the requested page). `total_count` is omitted when an `issue` or `sub_product` filter is active (those apply after the upstream query, so the upstream total wouldn't match). If CFPB is unreachable the tool falls back to a small cached sample (source:'cache' + coverage_warning) — do NOT infer volume in that mode. Note: `company` matching is word-based against the company name ('experian', 'wells fargo'), not arbitrary-substring. Product taxonomy (the top categories): - 'Credit reporting or other personal consumer reports' — Equifax, Experian, TransUnion. ~80% of recent complaint volume. - 'Debt collection' - 'Mortgage' - 'Credit card or prepaid card' - 'Checking or savings account' - 'Payday loan, title loan, or personal loan' - 'Money transfer, virtual currency, or money service' - 'Vehicle loan or lease' - 'Student loan' Company-response values: 'Closed with explanation', 'Closed with non-monetary relief', 'Closed with monetary relief', 'In progress', 'Untimely response', 'Closed without relief'. Cross-source tip: pair with get_enforcement_actions(source:'cftc'|'occ'| 'fdic'|'sec'|'doj', text:'<company>') to see if complaint volume preceded a formal enforcement action.

Returns US patent applications from the USPTO Open Data Portal, keyed on the APPLICANT (the corporate owner). Each record is one application's front-page metadata: title, applicant(s), first inventor + inventor count, filing/effective dates, status, entity size, application type. Follow source_url (USPTO Patent Center) for the full file wrapper / documents. Use this when the user asks: what is a company patenting, how many patents did a company file (and when), recent patents in a company's portfolio, or to cross-reference R&D output against insider/congressional activity, contracts, or fundamentals for the same company. company_name is the corporate APPLICANT and matches case-insensitively, but patents are filed under an IP-HOLDING ENTITY, not the household brand. Use the full legal applicant string for best recall — e.g. 'Google LLC', 'Microsoft Technology Licensing, LLC', 'Amazon Technologies, Inc.', 'QUALCOMM Incorporated', 'International Business Machines Corporation', 'Meta Platforms, Inc.', 'NVIDIA Corporation', 'Lockheed Martin Corporation', 'The Boeing Company', 'Apple Inc.', 'Intel Corporation', 'Tesla, Inc.'. COVERAGE — live-first (source:'live'): when company_name is set, each call queries USPTO live over the full ~12.9M-application index (the whole filing history of that applicant, most-recent first), with since/until on filing_date applied at the source. On USPTO outage the tool falls back to a cached recent slice for a dozen tracked big filers (source:'cache' + coverage_warning) — don't infer totals there. With NO company_name it serves that same recent cache (browse). application_number is a direct lookup.

Returns SEC Form C filings — Regulation Crowdfunding offerings, 2016-05→present: startup raises on Wefunder / StartEngine / Republic and other funding portals. One record per filing with the issuer (legal form, jurisdiction, incorporation date, website), the PORTAL (name + CIK + CRD), offering terms (security type — SAFEs appear as 'Other' with the description, price, target and maximum amounts, deadline, oversubscription), and the issuer's own DISCLOSED FINANCIALS (total assets, cash, revenue, net income, debt — current + prior fiscal year, dollars) plus employee count. Use this when the user asks about: startup crowdfunding activity, what a company raised on a portal, portal market share, early-stage issuers in a state, or revenue/assets of a crowdfunding company. filing_type maps the form family: 'offering' (C, C/A) | 'progress_update' (C-U) | 'annual_report' (C-AR — re-discloses financials yearly) | 'termination' (C-TR). is_withdrawal covers the -W variants. One issuer CIK typically has a chain: C → C-U → C-AR… — filter cik + sort asc to read it. Financial fields are null where a variant omits them. Reg A+ (Form 1-A 'mini-IPOs') is a different form family — planned as its own dataset; Reg D private placements are in get_private_placements. Pure-publisher posture: issuer-reported numbers as filed, parsed by KeyVex — Form C financials are self-reported and generally unaudited (reviewed at most); treat them accordingly.

Returns daily end-of-day closing-price history for one US-listed ticker (stocks, ETFs, mutual funds — including delisted tickers, so historical analysis is survivorship-bias-free). Coverage extends back as far as 1962 for the oldest names, subject to plan history limits. PAID PLANS ONLY. Each row: date, close (as-traded), adj_close (split+dividend adjusted — use THIS for charts and return calculations), div_cash (dividend with that ex-date), split_factor (e.g. 4 = 4:1 split that session). price_currency (top level) is the ISO currency of EVERY price in the response: USD. KeyVex covers US-listed securities priced in US dollars (foreign companies listed on US exchanges included); a ticker listed only elsewhere, e.g. a Shanghai or Shenzhen A-share, answers 'Not covered' with no prices. null means the currency is not known. include_ohlc=true adds the session's open / high / low / volume and their adjusted variants — the day's RANGE, which is what a stop or a target is actually tested against. On weekly/monthly these are aggregated over the period (first open, highest high, lowest low, summed volume), not the last session's values. Omit the flag and the response is unchanged. At most 5,000 rows per call (5,000 daily rows is about 20 years). When the window holds more, the response says truncated=true. Without since, you get the NEWEST 5,000 and prev_until: call again with until=prev_until for the page before. With since, you get the 5,000 from since onward and next_since: call again with since=next_since for the page after. Rows are never dropped silently. For multi-year ranges prefer frequency='weekly' or 'monthly' (last bar per period; dividends summed, split factors compounded), which never page: 10 years daily ≈ 2,500 rows vs ~120 monthly. One call = one ticker. Compare securities with multiple calls. Close Prices from Tiingo.com.

Returns SEC delisting and deregistration filings: the Form 25 family (notification of removal from listing on a national exchange under Rule 12d2-2) and the Form 15 family (certification terminating or suspending a security class's registration — the 'going dark' filing that ends SEC reporting; 15F variants are the foreign-private-issuer equivalents). Use this when the user asks: was/is a company being delisted, which companies went dark recently, what securities did an exchange remove, or to pair with tender offers / 8-Ks / insider sales around an exit event. Reading a record: action='delisting' (25 family) vs 'deregistration' (15 family) is a faithful form→rule mapping, not an opinion. 25-NSE is filed BY THE EXCHANGE against the issuer (exchange_name/exchange_cik are set) — typically the involuntary path; a bare Form 25 is filed by the issuer itself (voluntary withdrawal, e.g. after a merger). rule_provision carries the cited Rule 12d2-2 provision verbatim — the provision distinguishes the grounds for removal; agents can read the cited paragraph. A merger close typically produces a Form 25 AND a Form 15 within weeks. EDGAR coverage: Form 15 family 1994→present; issuer-filed Form 25 from mid-2001; exchange-filed 25-NSE from 2005 Q4 (the Rule 12d2-2 amendments moved exchange filings onto EDGAR — earlier removals were paper-filed and are not in EDGAR). security_class / rule_provision / exchange fields are populated from the structured 25-NSE XML; 15-family records are metadata-level — follow filing_index_url for the document. Pure-publisher posture: EDGAR records as published.

Returns FDA FAERS drug adverse-event reports — every adverse-event / medication-error report submitted to FDA (~20M, 2004→present, growing ~2M/yr). LIVE passthrough to openFDA: results reflect FDA's current data and `total_count` is openFDA's authoritative count for the filtered query (the results array is just the requested page). Use this when the user asks about: safety signals on a drug, adverse events by reaction type, death/hospitalization outcome counts for a product, a manufacturer's adverse-event footprint, or to pair a safety-signal trend with recalls, approvals, or insider activity. count_by returns TOP TERMS + COUNTS instead of records (e.g. count_by:'reaction' with drug:'ozempic' → the most-reported reactions for that drug) — the right first move for 'what are the side effects of X' questions; follow with a record query for detail. Records are openFDA's fields verbatim: deeply nested (patient.drug[] with openFDA annotations, patient.reaction[]), 5-15KB each — keep limit small. Matching: drug matches brand name, generic name, or the verbatim reported product as a PHRASE ('ozempic', 'semaglutide'); reaction is a MedDRA term phrase ('myocardial infarction'); serious=true filters to reports with a serious outcome; outcome picks one specific flag (death, hospitalization, …). since/until window on receivedate (when FDA received the report). CRITICAL honesty note (FDA's own): FAERS reports are UNVERIFIED and establish NEITHER causation NOR incidence — anyone can report, duplicates exist, reporting is stimulated by publicity, and there is no denominator (prescriptions dispensed). Counts are a reporting signal, not a risk measure. Surface this caveat when presenting counts. Pure-publisher posture: FDA's records as published, parsed by KeyVex — no derived safety scores.

Returns observations of key US macro, energy, and fiscal indicators from four sources: - BLS (Bureau of Labor Statistics): the canonical labor + price statistics. ~20-series watchlist covering unemployment, payrolls, wages, CPI, PPI, productivity. Most monthly, ECI/productivity quarterly. - FRED (Federal Reserve Economic Data, St Louis Fed): rates, money supply, GDP, PCE inflation, mortgage rates, jobless claims, Fed balance sheet, breakeven inflation, dollar index, consumer sentiment. ~30-series watchlist. Some daily (rates, dollar), weekly (mortgage, Fed assets, jobless claims), monthly, quarterly. - EIA (Energy Information Administration): WTI + Brent crude oil spot prices, Henry Hub natural gas, US gasoline retail price, US crude oil production. Unique energy data not in BLS or FRED. Mostly weekly cadence. - FiscalData (Treasury Bureau of the Fiscal Service): total public debt outstanding TO THE PENNY, daily, 1993→present (split into debt held by the public vs intragovernmental); Monthly Treasury Statement gross receipts / outlays / deficit-or-surplus; average interest rate actually paid on each Treasury security class (Bills/Notes/Bonds/TIPS/FRN + nonmarketable, 2001→present). Filter to one source via `source: 'bls' | 'fred' | 'eia' | 'fiscaldata'`. Default returns all four unified — `series_id` disambiguates across catalogs. Use this when the user asks about: unemployment rate, jobs report, nonfarm payrolls, CPI / PCE / inflation, Fed Funds rate, Treasury yields, mortgage rates, yield-curve inversion, money supply / M2, Fed balance sheet / QE / QT activity, GDP, housing starts, retail sales, consumer sentiment, jobless claims, trade balance, dollar strength, national debt / debt ceiling levels, monthly federal deficit, interest cost on the debt, or general macro context for cross-source analysis. Categories (for filtering): - rates — Fed Funds, Treasury yields, mortgage, corporate bonds - gdp — Real + nominal GDP, GDP growth rate - activity — Industrial production, housing starts, retail sales - inflation — CPI/PPI (BLS) + PCE/Core PCE/breakevens (FRED) - employment — Unemployment rates (U-3, U-6), payrolls, jobless claims - labor-force — Labor force participation rate - wages — Average hourly earnings, employment cost index - hours — Average weekly hours - productivity — Nonfarm productivity, unit labor costs - money — M2, Fed total assets, overnight reverse repo - debt — Federal debt (FRED quarterly + FiscalData daily to-the-penny), Treasury general account - fiscal — MTS monthly receipts, outlays, deficit/surplus (positive = deficit, negative = surplus, per Treasury's sign convention) - trade — Trade balance, trade-weighted dollar index - sentiment — U Michigan Consumer Sentiment - energy — WTI/Brent crude, Henry Hub natural gas, retail gasoline, US crude production (EIA) Period format is fixed-width per cadence so lexicographic sort = chronological: - 2026M04 (April 2026), 2026Q01 (Q1 2026), 2026A01 (annual 2026), 2026W18 (week 18 of 2026), 2026D258 (day-of-year 258). Set `latest_only=true` to get one record per series (the most-recent observation) — useful for 'where are things now' snapshot questions. Daily series under latest_only return only the latest day per series (deduped client-side); without it you can pull arbitrary history. Pure-publisher posture: the unit on each series is documented in the `unit` field; we do not compute year-over-year deltas, seasonally adjust differently, or derive 'real' vs 'nominal' versions — agents do those calculations on top.

Returns SEC + DOJ + CFTC + OCC + FDIC + FTC + Federal Reserve + FinCEN enforcement-related actions. Eight regulators, one tool. Use this when the user asks about: recent SEC charges, DOJ indictments, CFTC derivatives/swaps enforcement, OCC national-bank examination actions, FDIC bank-failure announcements or insured-deposit transfers, FTC antitrust / consumer-protection cases, Federal Reserve actions against banks and individual bankers, FinCEN anti-money-laundering (BSA) penalties, insider trading prosecutions, FCPA actions, fraud cases, or to add a 'negative event' flag to a ticker or person by cross-checking against insider trades, activist filings, or tender offers. CATEGORY: every row is labelled `category`: "enforcement" (charges, suits, settlements, penalties, consent and court orders, monthly enforcement-action lists; every DOJ, Federal Reserve and FinCEN row) or "agency_news" (rule proposals, no-action letters, appointments, roundtables, bank-failure notices, speeches from the SEC, CFTC, OCC, FDIC and FTC press feeds). By DEFAULT only enforcement is returned; pass category="all" for everything, or category="agency_news" for the news alone. A direct action_id lookup returns the row whatever its category. Sources: source='sec' — SEC press releases (sec.gov/news/pressreleases.rss). Rolling ~50-item RSS window; refreshes daily. SEC enforcement and policy statements are mixed in the same feed — filter by title substring (e.g., 'charges', 'fraud', 'insider trading') to narrow. source='doj' — DOJ press releases (justice.gov/api/v1/press_releases.json). Latest ~200 records refreshed daily; rich metadata including agency_component (issuing division) and topics[]. Common components: 'Criminal Division', 'Antitrust Division', 'Tax Division', 'Civil Division', 'Office of Public Affairs', 'United States Attorneys'. source='cftc' — CFTC press releases (cftc.gov/PressRoom/PressReleases). HTML index scrape (no RSS). Rolling ~50-item window. Covers derivatives/swaps enforcement, prediction-market jurisdiction, spoofing prosecutions, and policy actions. v1A index-only (no body extracted) — follow `url` for the substantive announcement. source='occ' — OCC news releases (occ.treas.gov/news-issuances/ news-releases/<year>/...). Covers national-bank enforcement, examination findings, capital/leverage rules, interagency announcements. Yearly index. Both OCC-only (`nr-occ-...`) and interagency (`nr-ia-...`) releases included; bulletins filtered out. source='fdic' — FDIC press releases (fdic.gov/news/press-releases). Covers bank failures + insured-deposit transfers, exam-result releases, deposit-insurance rule changes, CRA evaluations. Bank-failure announcements are some of the highest-signal FDIC items for agents. source='ftc' — FTC press releases (ftc.gov RSS). Antitrust, merger reviews, deceptive-practices and consumer-protection enforcement. Rolling recent window. source='fed' — Federal Reserve enforcement actions (full history from the Board's enforcement-actions CSV). One record per action per party: cease-and-desist orders, civil money penalties, prohibitions from banking, written agreements — against BOTH banking organizations and individual bankers. topics[] holds the action type(s); agency_component holds the bank (or the individual's affiliated bank); terminated_date is set once the Fed terminates the action (absent = still open). source='fincen'—FinCEN enforcement actions (fincen.gov, complete history). Rare, high-profile anti-money-laundering / Bank Secrecy Act penalties (e.g., TD Bank, Paxful, Brink's). topics[] holds the institution category; release_number holds the matter number; url points at the consent-order PDF. v1A scope: metadata + teaser + description (capped ~3000 chars; empty for CFTC v1A). Full prose lives at `url` — agents follow for the substantive announcement. Pure-publisher posture: no derived 'severity' or 'outcome prediction' signals. Identifier format: action_id is 'sec-{guid-or-slug}', 'doj-{uuid}', 'cftc-{release-number}', 'occ-{slug}', 'fdic-{slug}', 'ftc-{slug}', 'fed-{date}-{party}-{action}', or 'fincen-{matter-number}'. Stable across re-scrapes. Cross-source tip: pair with get_insider_transactions to detect insider trades by executives at companies later named in enforcement charges, or with get_activist_stakes to spot enforcement-driven exit attempts.

Returns EPA federal CIVIL enforcement cases from ICIS FE&C (the EPA's Integrated Compliance Information System) via the ECHO bulk download — ~135K cases, EPA-lead administrative and judicial civil actions. CRIMINAL prosecutions are NOT in this source, and neither are state-lead actions. Refreshed weekly by EPA (~Saturday). Use this when the user asks about: EPA fines / penalties against a company, Clean Air Act / Clean Water Act / RCRA / Superfund enforcement, environmental violations by facility or state, settlements and consent decrees, or supplemental environmental projects (SEPs). Record shape (one doc per case, joins pre-flattened): case_number (RR-YYYY-NNNN), case_name, defendants[] (names), statutes[] + primary_statute (CWA, CAA, FIFRA, SDWA, RCRA, TSCA, CERCLA, EPCRA), activity_type ('administrative' | 'judicial'), activity_status + status_date, penalties from the CASE_PENALTIES table — fed_penalty, state_local_penalty, sep_amount, compliance_action_cost, cost recoveries, penalty_collected — settlement_lodged_date (earliest; only ~34% of cases lodge, mostly judicial) + settlement_entered_date (latest) + settlements_count, facilities[] (name, city, state, NAICS, FRS registry ID), region_code, doj_docket_number, enf_outcome, voluntary_self_disclosure, multimedia, summary_text, and source_url (the ECHO case report page). Date filters (since/until) and the default sort use status_date — the case's last status-change date, present on every case. Sorting by fed_penalty cannot be combined with since/until (numeric field). Cross-source: pair with get_enforcement_actions (press-release actions from 8 other regulators), get_federal_contracts (whether an EPA defendant still wins federal awards), and get_material_events (8-K environmental-liability disclosures). Pure-publisher posture: EPA's case records as published — no derived severity scores or compliance opinions.

Returns FDA approval / clearance events: drug approval actions from Drugs@FDA, medical-device 510(k) clearances, and device PMA (premarket approval) decisions. Full history (drugs to 1939, 510(k) to 1976, PMA to the 1960s). This is the BULLISH twin of get_product_recalls — the catalyst dataset for biotech and medtech tickers. Use this when the user asks about: new drug approvals for a company or ingredient, priority-review approvals, tentative generic (ANDA) approvals, device clearances by company or product code, PMA supplements, or to pair an approval date with insider trades / 8-K filings / fundamentals. Sources (filter via the `source` enum): drugsfda — Drugs@FDA submission actions. One record per submission decision (ORIG = original approval, SUPPL = supplemental). decision_code: AP (approved) | TA (tentative approval — generic approved but blocked by patent/exclusivity). review_priority: PRIORITY | STANDARD — PRIORITY reviews are the higher-signal events. application_number prefix tells the product class: NDA (new drug), ANDA (generic), BLA (biologic). 510k — Device premarket notifications. decision_code SESE ('substantially equivalent' — cleared) dominates ~98%. approval_type: Traditional | Special | Abbreviated. pma — Device premarket approvals (Class III, highest-risk devices — implants, life-sustaining). Originals AND supplements (supplement_number, supplement_reason). decision_code: APPR (approved) | OK30 (30-day supplement accepted) dominate. openFDA ships no description text for PMA codes; decision_description mirrors the code. review_priority='EXPEDITED' marks devices under expedited review; 'PRIORITY' marks priority-review drugs. Empty = standard / not flagged. Company matching: `applicant` is a substring filter on the sponsor / applicant name AS FILED (e.g., 'Pfizer', 'Boston Scientific'). FDA records carry no ticker or CIK — subsidiaries file under their own names, so try the operating-company name, not the holding company. source_url points at the accessdata.fda.gov detail page (approval letters, labels, review documents). Pure-publisher posture: no derived 'approval odds' or price-impact signals.

Returns FEC-registered candidate profiles (House, Senate, President) and — when include_committees=true (default) — each candidate's associated FEC committees in the same response. Use this when the user asks about: who's running in race X, the campaign finance ID for a member, what PAC is sponsoring a candidate, or to bridge from a Congressional member name to their FEC committee_id before looking up contributions (v1.1 tool). Source: api.open.fec.gov — the official Federal Election Commission public-disclosure API. Records include current sitting members, primary challengers, defeated candidates, future-cycle registrants, and presidential candidates. Cycles tracked: 2022, 2024, 2026. Filter by candidate_id for the fastest direct lookup. Otherwise use candidate_name (case-insensitive substring) optionally narrowed by office + state + cycle. FEC names are typically filed as LASTNAME, FIRSTNAME (e.g., 'MCCORMICK, DAVE' for Dave McCormick). Office codes: H (House), S (Senate), P (President). Party codes: DEM (Democratic), REP (Republican), LIB (Libertarian), GRE (Green), IND (Independent), OTH (Other). Incumbent_challenge: I (Incumbent), C (Challenger), O (Open seat). When active_only=true, only candidates with candidate_status='C' (currently filing) are returned. Committee designations on returned committees: P (Principal campaign committee — the primary donation recipient), A (Authorized — accepts donations on candidate's behalf), B (Lobbyist), D (Leadership PAC), J (Joint fundraiser), U (Unauthorized). The Principal (P) committee is the one you want for 'donations to X's campaign'. Committee types: H (House campaign), S (Senate campaign), P (Presidential campaign), Q (PAC qualified), N (PAC non-qualified), O (Super PAC), I (Independent expenditure non-PAC), X/Y/Z (Party), V/W (Carey/hybrid). For 'follow the money to Senator X', look for designation=P, committee_type=S among the returned committees.

Returns FEC Schedule A contribution data — money flowing INTO federal committees — in AGGREGATED form. Individual donors are never exposed as searchable per-record rows: the FEC sale-or-use rule (11 CFR 104.15) permits aggregated presentation only, so this tool serves group totals and a bounded ORGANISATION leaderboard (the same posture as Quiver Quantitative's public pages). Source: api.open.fec.gov (official FEC API), queried live per request with a cached-rollup fallback (responses carry source: live | cache). THREE MODES (pick one): 1. Aggregate totals — pass group_by: - group_by='employer' + recipient_committee_id + cycle → total + count per employer for that committee (FEC-computed, all itemized rows). E.g. which employers' workforces fund committee X. - group_by='state' + recipient_committee_id + cycle → geographic fundraising pattern for a committee. ⚠ On employer and state rows, contribution_count is the number of CONTRIBUTIONS, not contributors — the FEC publishes no contributor count for these aggregates. A cell can carry a dozen contributions from ONE person (measured: an employer cell with 14 contributions and a single contributor). Do not read it as a crowd, and do not use it to judge whether a cell describes a population or an individual. - group_by='candidate' + cycle (optionally candidate_id) → per- candidate cycle receipts, itemized-individual share, disbursements, cash on hand. Sorted by receipts DESC — 'who raised the most'. - group_by='committee' + recipient_committee_id (cycle optional) → that committee's cycle totals. Top-committee LISTS come from the rollup cache and may lag a day. - group_by='cycle' + candidate_id or recipient_committee_id → per-cycle rows across cycles (fundraising trajectory). 2. Donor leaderboard — pass leaderboard=true + cycle + EXACTLY ONE scope: recipient_committee_id | candidate_id | contributor_state | contributor_employer. Returns top ORGANISATION donors (PAC / party / committee / company) as name + summed total + contribution count, PLUS the individual side as STATISTICS ONLY: individual_donor_count and individual_total. NO median and NO maximum are served — each is one person's number (a median over an odd count IS one contributor's gift) and re-identifies against the FEC's own site. ⚠ SMALL-CELL FLOOR: when fewer than 5 distinct individuals contributed in the scope, the whole individual block is WITHHELD — count and total both null, suppressed=true, and the envelope carries suppressed_small_cells. Three donors plus a total is three people's gifts nearly reconstructed, and the scope is public. The organisation board is never floored; entities are not natural persons. NO NATURAL PERSON IS NAMED BY THIS TOOL, IN ANY MODE. A paid service ranking named individuals by their contribution history is a prohibited commercial use of contributor lists (11 CFR 104.15; 52 U.S.C. 30111(a)(4)). An empty organisation list means no organisation gave in that scope above the floor — it is never a reason to look for people. No addresses, no city/ZIP, no per-record rows. The response's leaderboard.complete flag is true when every itemized row at or above the fixed $1,000 floor (min_amount is not accepted in this mode) for the scope was aggregated — totals are then exact; otherwise the pull hit its page cap and ranks are amount-weighted approximations. 3. Per-record (NON-INDIVIDUAL only) — pass entity_type (COM, CCM, PAC, PTY, ORG) with optional recipient_committee_id / candidate_id / contributor_state / cycle / amount / date filters, or sub_id for a direct lookup. Individual (IND) rows are never returned per-record; memo subtotals are excluded by default (exclude_memos=false to include). Useful for PAC-to-PAC transfer analysis. There is NO contributor_name search and no individual street/city/ZIP anywhere in this tool's output — by design, permanently. Killer query patterns: - Who raised the most this cycle? group_by='candidate' + cycle=2026. - Who funds Senator X? get_fec_candidate_profile → principal committee → leaderboard=true + recipient_committee_id + cycle. - Which employers' staff fund committee Y? group_by='employer' + recipient_committee_id + cycle. - Where does committee Y's money come from? group_by='state' + recipient_committee_id + cycle. - PAC-to-PAC flows into committee Z? entity_type='PAC' + recipient_committee_id.

Returns FEC Schedule B disbursements — itemized records of money flowing OUT of a federal committee to organisations: vendor payments, media / ad buys, consulting firms, payroll services, and committee-to-committee transfers. The OUT-flow counterpart to get_fec_contributions (Schedule A, money IN). ORGANISATIONS ONLY: per-record rows are served only when FEC codes the payee as a committee or organisation (COM, CCM, PAC, PTY, ORG). Rows naming a natural person — individual payees (IND), candidates (CAN), unclassified payees, and people a filer coded as an organisation — are withheld per record under the FEC sale-or-use rule (11 CFR 104.15). Refunds of contributions to individuals are withheld with them. Source: api.open.fec.gov/v1/schedules/schedule_b/ — the official FEC public-disclosure API. Live queries cover every itemized row; the cached fallback subset carries a $1,000+ ingestion floor (filters small-vendor / payroll noise). Publication-lag caveat: disbursements only surface when the spending committee FILES its report — monthly filers lag ~20 days, quarterly filers up to ~50 days after the spend. Short since/until windows on recent dates will miss rows whose reports haven't been filed yet. Killer query patterns: - What does candidate X's campaign spend money on? Pass spender_committee_id (their principal campaign committee from get_fec_candidate_profile). Sort by amount DESC for big spends. - Which campaigns pay consulting firm Y? Pass recipient_name='Y'. - Ad-spending patterns: disbursement_purpose_category='ADVERTISING' + cycle=2026 + sort_by=disbursement_amount DESC. - Money moving between committees: disbursement_purpose_category= 'TRANSFERS' — recipient_committee_id on each row names the receiving committee. Filter combinations note: server-side indexes support one equality filter (spender_committee_id / candidate_id / disbursement_purpose_category / recipient_state) combined with date or amount sort + cycle. Other filters (recipient_name substring, entity_type, exclude_memos) are applied client-side after a wider pre-fetch. Purpose categories (disbursement_purpose_category): ADVERTISING, CONSULTING, CONTRIBUTIONS, FUNDRAISING, PAYROLL (labeled 'SALARIES' on some rows), TRANSFERS, TRAVEL, ADMINISTRATIVE, MATERIALS, EVENTS, LOANS, REFUNDS, POLLING, OTHER. Memo rows: FEC tags certain aggregate / subtotal rows with memoed_subtotal=true. These DUPLICATE dollars already counted on other rows — KeyVex DEFAULTS to exclude_memos=true to show real money movement; pass exclude_memos=false to include the raw memo rows (e.g. for matching FEC's own row counts).

Returns FEC Schedule E independent expenditures — money spent BY a super PAC (or IE-only PAC) uncoordinatedly FOR or AGAINST a federal candidate. Hallmark vehicle for political ad spending since Citizens United (2010). Critical signal: support_oppose_indicator — 'S' = support, 'O' = oppose. A single candidate often has dozens of S and O entries across many super PACs in one cycle. Filter by support_oppose='O' to find attack ads; 'S' to find positive ads. Source: api.open.fec.gov/v1/schedules/schedule_e/. F24 filings (24-hour notices within 20 days of an election) and F5 (quarterly IE reports) both flow through this endpoint. Filers FEC classes as "a person or a group" (Form 5 filers, committee types I and E) are served when the filer is a group; a filer whose name is, or may be, a natural person's is withheld per record, as is any filer KeyVex holds no committee record for (FEC sale-or-use rule, 11 CFR 104.15). Killer query patterns: - Attack ads on Senator X: candidate_id='S6PA00091' + support_oppose='O' - Recent spend across a cycle: support_oppose='O' + since='2025-01-01' - Top political ad vendors: payee_name='AXIOM' (substring) - Negative spend in a race: candidate_office_state='PA' + support_oppose='O' Filter combinations note: server-side indexes support one equality filter (candidate_id / committee_id / support_oppose / candidate_office_state) + a date / amount sort. Other filters (payee_name, description, exclude_memos) are applied client-side. NOTE on cycle: the FEC leaves two_year_transaction_period null on many rows, so the `cycle` filter is INCOMPLETE (it undercounts) and is not currently indexed — scope a cycle by DATE instead, e.g. since='2025-01-01' for the 2026 cycle.

Returns federal contract awards from USAspending.gov — government spending data sourced from Treasury/GSA. Each record is one prime contract award (BPA Call, Purchase Order, Delivery Order, or Definitive Contract). Modifications appear as separate records. Use this when the user asks about: who's getting federal contracts, how much a specific recipient (Lockheed Martin, RTX, Raytheon, Booz Allen, etc.) won this year/quarter, contracts by industry (NAICS code) or product type (PSC code), or to cross-reference congressional trading with contract awards. Cross-source pattern (the political-alpha play): 1. get_congressional_trades(ticker:'LMT', since:'2026-01-01') — find LMT trades by members of Congress. 2. get_federal_contracts(recipient_name:'Lockheed Martin', since:'2026-01-01') — find LMT contract awards. 3. Compare timing — trades within 30 days before a major contract are the high-signal cases. ⚠ award_amount and total_outlays are NULLABLE, and total_outlays is null on MOST rows. USAspending omits Total Outlays from the search response about 75% of the time, and this tool now reports that as null rather than as $0 — a 0 here means the source really said zero. Do not do arithmetic on either field without a null check. Rows with a null value are excluded from min_amount filters and sort LAST, because a value we do not have cannot satisfy a threshold. recipient_name is a case-insensitive substring match — use the parent name ('Lockheed Martin') to catch all subsidiaries. COVERAGE — live passthrough (source:'live'): each call queries USAspending's API over the full dataset (2007-10 onward), with recipient/NAICS/PSC/min-amount/date filters applied server-side. The response's `total_count` is USAspending's authoritative award count for your filtered query — USE IT for volume answers (the `results` array is just the requested page). `total_count` is omitted — and coverage_warning says why — when USAspending cannot count the answer: a recipient_uei that is also the PARENT of other recipients (USAspending cannot filter to one UEI's own awards), recipient_name combined with recipient_uei, or a start_date window (sort_by start_date with since/until). Then has_more is true whenever the search stopped before the end. On USAspending outage the tool falls back to a recent cached window (source:'cache' + coverage_warning) — don't infer volume there.

Returns federal GRANTS and cooperative agreements from USAspending. Distinct universe from get_federal_contracts — recipients here are universities, non-profits, state and local agencies, research labs, healthcare institutions, public-private partnerships. ⚠ award_amount and total_outlays are NULLABLE. USAspending omits Total Outlays from the search response for most grants, and this tool reports that as null rather than as $0 — a 0 means the source really said zero. Null-check before doing arithmetic. Null values never satisfy min_amount and sort last. Award type codes covered: 02 (Block Grant), 03 (Formula Grant), 04 (Project Grant — most common), 05 (Cooperative Agreement). Killer query patterns: - All NIH R01 grants this quarter: cfda_number='93.847' + since=... - State and local infrastructure funding: awarding_agency='Department of Transportation' + min_amount=1000000 - Recipient-specific grant history: recipient_name='Stanford' - Recipient by federal UEI: recipient_uei='ABC123XYZ' (most precise) Source: api.usaspending.gov — official Treasury federal-spending data. Awards covering both COVID/IIJA emergency-funding codes and routine appropriations. Pure-publisher posture: raw award data, no derived rankings or performance scores. COVERAGE — live passthrough (source:'live'): each call queries USAspending's API over the full dataset (2007-10 onward), with recipient/CFDA/min-amount/date filters applied server-side. The response's `total_count` is USAspending's authoritative grant count for your filtered query — USE IT for volume answers (the `results` array is just the requested page). `total_count` is omitted — and coverage_warning says why — when USAspending cannot count the answer: a recipient_uei that is also the PARENT of other recipients (USAspending cannot filter to one UEI's own grants), recipient_name combined with recipient_uei, or a start_date window (sort_by start_date with since/until). Then has_more is true whenever the search stopped before the end. Note: live cfda_number matching is against the award's FULL assistance-listings array (awards can carry several CFDAs); the cached fallback matches the primary listing only. On USAspending outage the tool falls back to a recent cached window (source:'cache' + coverage_warning) — don't infer volume there.

Returns Federal Register documents — the daily-published collection of US executive branch regulatory + administrative actions. Use this for: regulatory tracking (what's the SEC / EPA / FDA proposing this week?), executive order monitoring, public-comment-period tracking, lobbying tie-in (cross-reference with get_lobbying_filings for 'who's pushing which rule'), or compliance forward-look on proposed regulations. Source: federalregister.gov public REST API. Comprehensive — every Federal Register publication appears here. Document types (document_type field): 'Rule' — final regulation (in effect) 'Proposed Rule' — agency rule open for public comment 'Notice' — formal notice (sunshine acts, hearings, authorizations, determinations, etc.) 'Presidential Document' — executive orders, proclamations, memoranda Agency filtering: agency_slug uses URL-safe identifiers like 'securities-and-exchange-commission', 'environmental-protection-agency', 'food-and-drug-administration', 'federal-trade-commission'. Filter via agency_slug for exact-match; agency_name does case-insensitive substring for fuzzier 'I know the name but not the slug' lookups. Composite document numbers (e.g., '2026-09385') are GPO-assigned and stable. Direct document_number lookup is fastest. Pure-publisher posture: KeyVex returns the daily publication record as-is. No 'regulatory risk score' or 'likely-to-finalize' signals.

Returns federal disaster declarations from OpenFEMA — every DR (major disaster), EM (emergency), and FM (fire management) declaration since 1953, one record per (declaration, designated county). ~70K records. Use this when the user asks about: hurricanes / floods / wildfires / severe storms hitting a state or county, which counties were designated for FEMA assistance, active vs closed-out disasters, or to anchor an insurance / construction / utility / muni-credit question to the official federal declaration. Record shape: fema_declaration_string ('DR-4728-CA'), declaration type + date, incident_type ('Hurricane', 'Flood', 'Fire', 'Severe Storm'...), declaration_title ('HURRICANE IAN'), designated_area (county) + FIPS codes, and the four assistance-program flags (individual_assistance, individuals_households_program, public_assistance, hazard_mitigation) — public_assistance=true is the infrastructure-rebuild-money flag. A single disaster spans MANY records (one per designated county): filter fema_declaration_string or disaster_number for all areas of one event; filter state + since for a state's recent disasters. Cross-source: follow a declaration with get_federal_contracts / get_federal_grants (recipient_name or date-windowed) for the rebuild spend, and get_material_events for insurer 8-Ks after major events. Pure-publisher posture: FEMA's declarations as published — no derived damage estimates or exposure scores.

Returns FERC eLibrary document records — every public filing at the Federal Energy Regulatory Commission: Issuances (orders, notices, delegated letters BY FERC) and Submittals (rate filings, tariff changes, compliance reports, protests, hydro license paperwork TO FERC) across the Electric, Natural Gas, Oil, Hydro, Rulemaking, and General libraries (~1,300 documents/week). Use this when the user asks about: a FERC docket or rate case, pipeline / utility / hydro regulatory activity, FERC orders affecting a company, or energy infrastructure proceedings. The DOCKET NUMBER is the join key across a proceeding. Format: PREFIX + two-digit year + sequence, e.g. 'ER26-1234' (Electric rate), 'RP26-930' (gas pipeline rate), 'CP26-15' (gas pipeline certificate/construction), 'P-5737' (hydro project — no year), 'EL26-50' (Electric complaint/investigation), 'RM26-3' (rulemaking). docket_number accepts the root ('RP26-930') or a full sub-docket ('RP26-930-000'). Records carry docket_numbers (verbatim) plus docket_prefixes for quick classification. Record shape: accession_number (e.g. '20260708-3001' — the unique document key), category ('Issuance' | 'Submittal'), description, filed/issued/posted dates, class_types (document class + type), libraries, authors[] / recipients[], and transmittal file metadata. Documents themselves are NOT in the record — agents download the primary attached file via source_url (eLibrary filedownload link). Pure-publisher posture: FERC's search metadata as published — no outcome predictions or case scoring.

Returns FARA registrations — US persons and firms registered with the DOJ as agents of a foreign principal under the Foreign Agents Registration Act. Use this when the user asks about: who is a registered foreign agent, which US firms work for a particular foreign government, recently-registered foreign agents, or to add a 'foreign- influence' flag to a lobbying firm, law firm, or PR firm. Each record is one registrant ↔ foreign-principal relationship — a registrant representing three foreign principals appears as three records. The single highest-signal filter is foreign_principal_country: foreign_principal_country='CHINA' → every US agent acting for a Chinese principal Source: efile.fara.gov (DOJ National Security Division). v1A covers ACTIVE registrations. The registrant↔principal linkage is included; per-document filing detail and compensation figures are not — follow source_url to FARA eFile for those. Cross-source pairing pattern: FARA + get_lobbying_filings — FARA is foreign-principal representation; LDA is domestic lobbying. A firm in both is lobbying Congress on behalf of a foreign government. FARA + get_fec_contributions — foreign-agent firms whose people also make political contributions. FARA + get_congressional_trades — influence-and-trades overlay. Identifier: registration_number is the FARA registration number. has_foreign_principal=false records are registrants with no currently- active foreign principal (still queryable as registered agents). History: registrations that LEAVE DOJ's active list are kept with status:'terminated' (+ termination_observed_date) rather than deleted — a terminated registration is still real history. Default queries return BOTH; filter status:'active' for the current roster only.

Returns per-security holdings from SEC Form N-PORT primary documents — one row per investment-or-security line in a mutual fund / ETF / closed-end fund's monthly portfolio report. Use this when the user asks about: which funds hold a specific stock or bond, a fund's complete portfolio composition, fund-level derivative exposure (swaps, options, futures), repo positions, concentration by issuer, or to compose 'which ETFs added X this month' / 'which funds shorted Y' style queries. Source: parsed from each NportFiling's primary_doc.xml. asset_cat is SEC's own code, and these twenty are the only ones the data holds — EC common equity; EP preferred equity; DBT debt, Treasuries included; LON loan; ABS-MBS asset-backed, mortgage; ABS-O asset-backed, other; ABS-CBDO asset-backed, CDO/CBO; ABS-APCP asset-backed commercial paper; SN structured note; STIV short-term investment vehicle, money-market and liquidity pools; RA repurchase and reverse-repurchase agreement; RE real estate; COMM commodity; DE equity derivative; DIR interest-rate derivative; DFE foreign-exchange derivative; DCR credit derivative; DCO commodity derivative; DO other derivative; OTHER other. Anything else is refused with this list. Useful filter combos: ticker='NVDA' all funds holding NVDA cusip='037833100' AAPL by CUSIP (more reliable than ticker for N-PORT) is_derivative=true, filer_name='BlackRock' the BlackRock FAMILY's derivative book — a name is a substring, and this one spans five trusts. filer_cik picks one of them. derivative_type='swap' every swap position in the universe asset_cat='RA' repo and reverse-repo exposure payoff_profile='Short' short positions only min_pct_of_portfolio=5 concentrated positions (>=5% of NAV) filer_cik='0000884394' one fund's complete portfolio counterparty='Morgan Stanley' every fund's exposure to one dealer min_notional=10000000 derivatives with >=$10M of EXPOSURE (not fair value — see min_notional) derivative_type='swap', expiration_until='2026-12-31' swaps rolling off before year-end maturity_since='2026-01-01', maturity_until='2026-12-31' bonds maturing this year Each holding ties to its parent NportFiling via filing_id. Read the parent for fund metadata (file_date, file_number, amendment flag); read the holding for security-level detail. is_derivative + derivative_type distinguish structured derivative rows from straight equity/debt holdings. DERIVATIVE + DEBT TERMS are extracted and filterable (2026-08-09). Derivatives carry deriv_notional_amount, deriv_counterparty_name/_lei, deriv_expiration_date, deriv_unrealized_appreciation, and the leg-level detail — put/call, written/purchased, strike, share count and delta for options; swap flag and upfront payment/receipt for swaps; both currency legs for forwards. Debt rows carry debt_maturity_date, coupon kind, annualized rate, default, arrears and paid-in-kind flags. USE min_notional, NOT min_value_usd, TO SIZE A DERIVATIVE. Fair value and exposure differ by orders of magnitude: a real interest-rate swap here shows value_usd -49,184 against a notional of 6,795,000. Ranking derivatives by value_usd understates them by ~100x. BACKFILL IN PROGRESS: these fields are populated on filings re-extracted since 2026-08-09 and are absent (null) on the rest until it completes. A null term on an older filing means not-yet-extracted, NOT absent from the source — do not read it as 'this swap has no counterparty'. The parent filing's primary_document_url always has the authoritative detail. COVERAGE: holdings begin with N-PORT filings FILED 2024-06-01. Most filings from then on have holdings rows, but not all: some are still metadata-only in get_nport_filings (not yet extracted — about 4% as of September 2026). Filings before 2024-06-01 are outside this coverage. For a filing with no holdings here, follow primary_document_url in get_nport_filings. An empty result for a filing means not-extracted, not an empty fund.

Returns XBRL-tagged financial fundamentals from public-company 10-K and 10-Q filings, sourced from SEC EDGAR's company-facts API. Each record is one observation of one concept at one period end. Use this when the user asks about: revenue, profit, margins, cash position, debt, shareholder equity, EPS, share count, operating vs. financing cash flow, or any line-item-level financial state of a public company. v1A scope: a curated 40-concept watchlist covering: - income_statement: Revenues / RevenueFromContractWithCustomer / CostOfRevenue / GrossProfit / OperatingExpenses / R&D / SG&A / OperatingIncomeLoss / InterestExpense / IncomeTaxExpenseBenefit / NetIncomeLoss - balance_sheet: Assets / AssetsCurrent / Cash / AccountsReceivable / Inventory / PP&E / Goodwill / Liabilities / LongTermDebt / StockholdersEquity / CommonStockSharesOutstanding - cash_flow: NetCash{Operating/Investing/Financing}Activities / PaymentsToAcquirePPE (capex) / PaymentsForRepurchaseOfCommonStock / PaymentsOfDividends / DepreciationDepletionAndAmortization - metrics: EarningsPerShareBasic/Diluted, weighted-avg share counts - entity: EntityCommonStockSharesOutstanding (dei taxonomy) Key cautions on the data: - The same concept can appear in multiple units (e.g., 'USD' and 'USD/shares' for EPS). Filter by unit if you need a specific shape. - Many concepts have BOTH year-to-date cumulative observations AND quarterly-period observations on 10-Q filings. The `frame` field (e.g., 'CY2025Q3') marks the per-quarter point-period observation; rows with empty `frame` are typically cumulative YTD. - Older filings may use deprecated concept names; KeyVex catalog includes both modern and legacy names where companies migrated (e.g., Revenues AND RevenueFromContractWithCustomerExcludingAssessedTax). Set `latest_only=true` to get one record per (ticker × concept) — the most-recent observation. Useful for 'current state' snapshots. Pure-publisher posture: values are AS FILED. We do NOT compute derived ratios (P/E, ROE, ROIC), YoY/QoQ deltas, or 'real' vs nominal versions. Agents calculate those on top.

Returns recent congressional + oversight publications from GovInfo across four collections. Use this when the user asks about: - Committee reports on a specific bill or topic - Recently signed public laws (the 'did it become law' signal) - Congressional hearing transcripts (testimony from regulators, CEOs, expert witnesses) - GAO oversight reports (independent reviews of federal agencies + programs, often precede SEC/DOJ enforcement on the same target) Collections (filter via the `collection` enum): CRPT — Congressional Reports. Includes committee reports accompanying bills (House hrpt / Senate srpt). Real- time signal on what's about to move on the floor. PLAW — Public + Private Laws. Bills that were signed into law. The 'what actually got done' record. CHRG — Congressional Hearings. Transcripts of House + Senate committee hearings — testimony from agency heads, executives, expert witnesses. Hearings often PRECEDE enforcement actions (the public 'why did this happen' conversation). GAOREPORTS — GAO oversight reports. Independent congressional oversight. NOTE: GovInfo's GAO collection is a historical archive (~16.5K reports) that is not receiving recent updates — GAO now publishes current reports on gao.gov directly. Use this for historical GAO research; recent reports won't appear here. Identifier format: each `package_id` is GovInfo's globally-unique ID (e.g., 'CRPT-119hrpt27' for House Report 27 of the 119th Congress, 'PLAW-119publ12' for Public Law 12, 'CHRG-119hhrg54321' for House hearing 54321). Direct doc lookup by package_id is fastest. Cross-source pairing pattern: Hearing → trade by attending member: get_congressional_trades( bioguide_id:'...', since:'<hearing date>') GAO report on agency → SEC follow-on: get_enforcement_actions( text:'<agency name>', since:'<GAO report date>') Committee report → bill passage: get_bills + get_roll_call_votes Full document body (PDF / HTML / XML) lives at `package_link`; v1A returns only metadata — agents follow the link for content.

Returns H-1B Labor Condition Applications from the Department of Labor's quarterly disclosure files — one record per LCA with employer, job title, O*NET-SOC occupation code, offered wage vs DOL prevailing wage, worksite location, and employer risk flags (h1b_dependent, willful_violator). Covers H-1B, H-1B1 (Chile / Singapore), and E-3 (Australia) visa classes. Use this when the user asks about: a company's hiring activity or wage levels for specific roles, tech-hiring trends by state or occupation, offered vs prevailing wage gaps, or outsourcing-firm staffing patterns. IMPORTANT interpretation notes (stated so agents don't over-read): an LCA is filed BEFORE the H-1B petition and can cover multiple positions (total_worker_positions) — it signals hiring INTENT, not an approved visa or a hire. Certified ≫ actual visas issued. Wage fields are as filed; wage_unit varies (Year / Hour / Month / Week / Bi-Weekly) — normalize before comparing. Matching: employer_name is a substring over legal name + DBA (e.g., 'infosys', 'amazon'). soc_code is the precise occupation filter (e.g., '15-1252.00' Software Developers). case_status values: 'Certified', 'Certified - Withdrawn', 'Denied', 'Withdrawn'. Coverage: FY2024→present from DOL's quarterly files (fiscal_year / fiscal_quarter on each record; ~400-550K filings per quarter). Pure-publisher posture: DOL's disclosure rows as filed — no derived 'real wage' normalization or employer scoring.

A ticker search also returns rows this issuer FILED UNDER SYMBOLS IT NO LONGER LISTS (renames, filer typos, ADR spellings). Each row keeps `ticker` as SEC received it and gains `current_ticker` when the issuer trades under a different symbol today. Separately-listed share classes are NOT merged, and asking for a RETIRED symbol returns only rows filed under it — retired symbols get reissued to other companies. Returns FILING-LEVEL records from SEC Form 3/4/5 filings — one row per filing (accession). Use this when the user asks: list an issuer's insider filings, find a specific accession, filter by form type (Form 4 trades vs Form 3 initial statements vs Form 5 annual vs their /A amendments), count filings over a period, or size a filing (how many transaction / holding rows it has) before pulling detail. This is the filing INDEX. For the actual trades use get_insider_transactions; for the positions use get_insider_holdings. They join on accession_number. Source: SEC bulk Form 3/4/5 dataset (insider_filings_v2), 2006→present, refreshed quarterly. Each row carries the SUBMISSION envelope (company_cik, company_name, ticker, document_type, filing_date, period_of_report, is_amendment), ALL reporting owners (reporting_owners[]), signature rows, and per-table counts (nonderiv_trans_count, deriv_trans_count, nonderiv_holding_count, deriv_holding_count, footnote_count). Useful filter combos: ticker='AAPL', document_type='4' all Apple Form 4 filings company_cik='0000320193', is_amendment=true Apple's amended filings only accession_number='0000320193-26-000078' one specific filing's envelope ticker='TSLA', since='2025-01-01' TSLA insider filings this year ticker='NVDA', reporting_owner_name='Huang' NVDA filings involving Huang document_type values are the raw SEC form codes: '3', '4', '5' and their amendments '3/A', '4/A', '5/A'.

A ticker search also returns rows this issuer FILED UNDER SYMBOLS IT NO LONGER LISTS (renames, filer typos, ADR spellings). Each row keeps `ticker` as SEC received it and gains `current_ticker` when the issuer trades under a different symbol today. Separately-listed share classes are NOT merged, and asking for a RETIRED symbol returns only rows filed under it — retired symbols get reissued to other companies. Returns per-security INSIDER POSITIONS from SEC Form 3/4/5 filings — one row per holding line reported by a corporate insider (director, officer, or 10%+ beneficial owner). Use this when the user asks: what a specific insider currently HOLDS, who the largest insider holders of a stock are, an insider's position across companies, or direct-vs-indirect ownership structure. This is the position (stock) companion to get_insider_transactions (the buys/sells flow). Source: SEC bulk Form 3/4/5 dataset (insider_holdings_v2), 2006→present, refreshed quarterly. Each row carries the filing envelope (company_cik, company_name, ticker, reporting_owner_cik/name, role flags, filing_date, period_of_report), the position (holding_type nonderiv|deriv, security_title, shrs_owned_following_trans, valu_owned_following_trans, direct_indirect_ownership D|I), and — for derivative holdings — conv_exercise_price, exercise_date, expiration_date, and underlying-security detail. Useful filter combos: ticker='NVDA', sort_order='desc' most-recent NVDA insider holdings reporting_owner_cik='0001214128' one insider's positions everywhere ticker='AAPL', holding_type='deriv' AAPL insiders' option/RSU positions ticker='TSLA', is_ten_percent_owner=true 10%+ owners of TSLA company_cik='0000320193', min_value=1000000 Apple insiders holding >$1M reporting_owner_name='Musk' name substring (case-insensitive) shares/value are 'following the reported transaction' — the position as of that filing, not a live real-time holding. For the trades themselves use get_insider_transactions; ownership ties together via reporting_owner_cik.

Returns executive insider transactions filed on SEC Form 4 — open-market purchases and sales by officers, directors, and 10%-owners of public companies. Each record is one transaction line item from one filing. Use this when the user asks about: insider buying or selling at a specific company, all recent insider activity across the market, transactions by a specific officer, or large insider trades by value. Form 4 is the fastest insider-trade signal in the public record — must be filed within 2 business days of the trade. The reporting_lag_days field tells you how stale a particular disclosure is. Returns BOTH non-derivative rows (direct common-stock buys/sells, RSU vests, grants, gifts, tax-withholding sales) AND derivative rows (option exercises, warrant conversions, RSU/PSU activity). Filter to one or the other with is_derivative; filter to specific transaction codes with transaction_codes. Common transaction codes: P open-market purchase | S open-market sale A grant / award / RSU vest | M exercise of derivative X exercise of in/at-the-money derivative | C conversion of derivative F payment of exercise price or tax with shares | G bona fide gift D disposition to issuer (forced) | I 401(k)/ESPP | V voluntary ⚠ shares and price_per_share are NULLABLE, and null does not mean zero. SEC permits either to be omitted — the price can live in a footnote, and the share count is genuinely undetermined on instruments that convert at a future price (a convertible note settling on a later VWAP). Those filings state a dollar amount instead, so such rows carry total_value with a null shares. Before 2026-08-18 they were dropped from this dataset entirely. Do not do arithmetic on either field without a null check, and do not read a null share count as a trade of nothing — read total_value. Useful filter combos: ⚠ transaction_codes=['P'] IS NOT 'open-market buys'. SEC defines P as 'open market OR PRIVATE purchase', and the code alone says nothing about whether the security is common stock. Verified 2026-08-14: FLUT's code-P rows are $250M of Total Return Swaps (is_derivative=true) and ATTO's are an $8.5M private placement. Both are correctly labelled in transaction_nature and security_title — but a screen filtered on the code alone ranks them top by size. transaction_codes=['P'], is_derivative=false, include_non_open_market=false genuine open-market common-stock buys — the combination you almost always want transaction_codes=['M','X'] option exercises (cash-out trigger) transaction_codes=['A'] grants / RSU vests is_derivative=true all option/RSU/warrant activity is_derivative=false, transaction_type='sell', min_value=1000000 large open-market sells of common stock Optional include_baseline=true: also returns matching Form 3 initial- ownership records (the insider's *starting* position when they first became an insider) under a `baselines` field. Use this when you need to know how big a sale is relative to the insider's full position — Form 4 alone shows the delta, Form 3 anchors the baseline. Requires ticker or company_cik to be set. Baseline rows with is_nil_filing=true are 'no securities owned' Form 3s (~half of all filings) — the insider filed but started with ZERO holdings; shares_owned 0 is the position, not missing data. A ticker search also returns rows this issuer FILED UNDER SYMBOLS IT NO LONGER LISTS — renames, filer typos and ADR spellings all split a company's history across symbols. Each row keeps `ticker` exactly as SEC received it and gains `current_ticker` when the issuer trades under a different symbol today; the response carries `ticker_resolution` naming every symbol searched. Separately-listed share classes are NOT merged: GOOG does not return GOOGL. Asking for a RETIRED symbol returns only rows filed under it, because retired symbols get reissued to other companies. Rows found under a retired symbol are checked against the issuer's CIK, so a symbol another company files under today cannot leak its rows in. Coverage: full history on the bulk leg; the live-feed leg is scanned back 180 days, so a rename in the last few days may not be covered yet. data_source SELECTS WHICH BACKING COLLECTION: 'bulk_v2' (DEFAULT as of 2026-05-24) — `insider_transactions_v2` collection populated by SEC quarterly bulk Forms 3/4/5 TSV bundles. Deeper history (2006q1 → latest published quarter, ~9.9M rows). ⚠ RECENCY: the bulk dataset ends at the last PUBLISHED quarter (SEC releases it ~2 weeks after quarter end). On simple recency queries (descending sort, no v2-only filters) filings newer than that boundary are AUTO-MERGED from the live daily feed, so the default view stays current — coverage_warning says when this happened. For post-boundary browsing with v2-only filters, query data_source:'legacy' directly. INLINED FOOTNOTES (footnote_refs[] with resolved text on every row), aff10b5one 10b5-1 plan flag, full reporting_owners array, schema_era. Filters: ticker, company_cik, reporting_owner_cik, reporting_owner_name (substring), row_type ('nonderiv'|'deriv'), trans_codes (aka transaction_codes — either spelling works on either data_source), aff10b5one, schema_era ('pre_2023'|'2023_plus'), since/until, sort_by ('transaction_date'|'filing_date'). PLAUSIBILITY FIELDS — WHAT THEY CAN AND CANNOT CONCLUDE: Every row carries price_check and volume_check, and both are ALWAYS non-null: they say whether each check ran, and why not when it did not. Read those first. ⚠ THE RAW MARKET VALUES ARE PAID-PLAN ONLY. price_range_low and price_range_high (bulk rows), daily_range (legacy and live-feed rows) and shares_vs_daily_volume are Tiingo market data, which KeyVex's licence restricts to paid plans. On other plans those keys are OMITTED — not null — and the response carries licensed_fields_withheld naming them. The verdicts computed from them (price_check, price_outside_daily_range, price_fits_date, volume_check, volume_verdict) are served on every plan. ⚠ THE OTHER THREE ARE OFTEN ABSENT OR NULL, AND THIS TEXT USED TO SAY 'every row carries' ALL FOUR, WHICH WAS FALSE. Measured 2026-09-04 by the KeyVex auditor over a 500-row market-wide March sample: price_outside_daily_range key present 500/500, NON-NULL on 178 volume_verdict key present 500/500, NON-NULL on 178 shares_vs_daily_volume key ABSENT on 500/500 The 322 nulls are exactly the rows where volume_check is not 'checked' — so the reason is always available, on the field that says so. shares_vs_daily_volume is stamped only alongside a NON-NORMAL volume verdict, and the same sample contained no non-normal rows, so an ordinary response carries it nowhere. Do not build on its presence. A null here means NOT CHECKED. It never means fine. They were served without definition until 2026-09-01, which is how 'impossible' came to read as a stronger claim than the check supports. volume_verdict compares REPORTED SHARES against that day's recorded volume, and shares_vs_daily_volume is the raw ratio so you can judge for yourself: 'normal' ratio <= 0.25 'outsized' ratio > 0.25 — a large share of the day's tape 'impossible' ratio > 1 — MORE SHARES THAN THE DAY RECORDED. ⚠ 'impossible' means the two numbers cannot both be right, NOT that the trade did not happen. Our volume is one daily bar: it need not include off-exchange or block prints, and a Form 4 may report several days' activity on one date. Treat it as strong evidence of a reporting or data problem worth investigating, not as proof the transaction is fake. null = not judged. Computed only for market claims (codes P and S) — a grant never touched the tape, so a ratio on it would be noise. price_check says whether the price was tested against that day's bar: 'checked' tested; price_outside_daily_range holds the result 'misdated' tested; the price is OUTSIDE that day's bar (see daily_range) — price_outside_daily_range is TRUE — but fits the bar of price_fits_date, within 3 calendar days. Read as: a real fill whose filer wrote the wrong date. KeyVex's own screens treat it as real: its dollar total is kept, it is not vetoed as non-open-market, and co-report resolution never drops it silently. 'no_price_reported' the filer stated no price 'no_positive_price' the filer stated a price of zero or less, so there was nothing to compare. The bar may be present — see volume_check. 'no_verdict_recorded' the day's bar was found (volume_check ran off it) but its high/low were unusable, so the price half is unjudged 'no_daily_bar' no usable bar for that ticker and date ⚠ THE LAST THREE ARE DELIBERATELY DISTINCT. Until 2026-09-04 all three were served as 'no_daily_bar', which asserted a missing bar on rows whose shares_vs_daily_volume — a ratio computable only FROM that day's bar — was served three fields away. Found live by the trading simulation. If you match on 'no_daily_bar', match on all four. volume_check says whether the SHARES-vs-VOLUME check ran, and is now independent of the price half: 'checked' | 'not_a_market_trade' | 'no_daily_bar'. A row can be volume_check 'checked' while price_check is 'no_positive_price' — the bar was there, only the price was not. price_outside_daily_range is true|false|null, and NULL MEANS NOT CHECKED — never 'fine'. A row whose price_check is any value other than 'checked' or 'misdated' has not been vetted on price at all, so do not read its silence as a pass. CLUSTER BUY (every data_source): cluster_buy_insiders_30d = the number of DISTINCT reporting owners (by CIK) with an open-market purchase (code P) in the same ticker in the 30 days ending on this row's transaction_date, this row included; cluster_buy = a code-P row with that count >= 3. Both are NULL — never a guess — on a row that is not a purchase, or when the window cannot be counted; cluster_buy_basis always says which, e.g. 'owner CIK not recorded for trades before 2026-07-01' (live-feed rows before then carry no owner CIK; bulk rows always do). BACKWARD-COMPAT: every v2 row also carries the LEGACY field aliases (disclosure_date, transaction_code, shares, price_per_share, total_value, acquired_disposed, shares_owned_after, officer_name, is_derivative, reporting_lag_days, data_source, sec_filing_url) so callers reading the old field names keep working. The `transaction_type` field carries the legacy 'buy'|'sell' semantic (synthesized from trans_code + trans_acquired_disp_cd, identical algorithm to the legacy scraper); the v2 nonderiv|deriv discriminator lives at `row_type`. 'legacy' — `insider_trades` collection populated by KeyVex's daily EDGAR scraper. Shallower coverage (2022+), no footnotes, no aff10b5one, ~91% fewer filings in the same window than bulk_v2. Rows written since 2026-09-30 also carry reporting_owner_cik (the filing's FIRST reporting owner, 10-digit, the bulk's own rule) and reporting_owner_ciks (every owner on the filing); older legacy rows do not, so the field's absence means 'not recorded', not 'none'. Filters: ticker, company_cik, officer_name, transaction_type (buy|sell), is_derivative, transaction_codes (aka trans_codes), min_value, since/until, sort_by (disclosure_date|transaction_date|total_value). Use this only when you specifically need the legacy doc shape with NO v2-extension fields. SEC-SOURCE DATE CONVENTIONS — read raw values with these in mind: KeyVex preserves SEC's authoritative bytes exactly as published. Two recurring source-data patterns are worth recognizing so agents interpret raw date values correctly: (1) PERPETUAL-INSTRUMENT SENTINEL — exercise_date or expiration_date values of 2050-12-31 / 2050-08-31 ARE SEC's established convention for instruments with no calendar expiration (Deferred Stock Units, certain Non-Qualified Stock Options, Units of Limited Partnership Interest, similar perpetual or condition-vested derivatives). Read these as 'no expiration,' not as literal calendar dates in 2050. This is a fact about SEC's schema, not an inference. (2) ANOMALOUS-YEAR FILER-ENTRY PATTERN — date values with out-of-range year components — e.g., 0012-11-21 or 0025-07-25 (likely 2-digit years entered into a 4-digit field), or 2027-01-25 on a 2026 filing / 2028-03-19 on a 2024 filing (likely single-digit transpositions) — appear to be filer data-entry typos preserved verbatim from SEC's primary filings. KeyVex verified on a stratified spot-check that these values are byte-identical between SEC's primary XML and SEC's bulk extract (22 / 22 matches across all observed pattern faces); the SEC-to-KeyVex transit is faithful. The pattern is ongoing — observed across filings from 2014 through 2026, not legacy-only. Cross-reference filing_date to infer the likely intended year. (3) NUMERIC PRECISION — for data_source='bulk_v2', shares and price_per_share mirror SEC's BULK Form 345 extract, which rounds to 2 decimals (e.g. 474.6, where the primary XML shows 474.598). That rounding is SEC's, in the bulk feed — KeyVex stores the bulk value verbatim (no rounding in the loader). Audit v2 numerics against the bulk extract (the source of record), not the XML primary document, which carries fuller precision. Dates and transaction codes DO match the XML exactly. (4) A DISCLOSURE DATE IS NOT CLOSED WHEN THE DAY ENDS. Filings keep arriving bearing a disclosure_date that has already passed, because SEC accepts them late and there is no cut-off after which a date stops gaining rows. So the SAME disclosure_date window can return MORE rows tomorrow than it did today, and a result cached against that date goes quietly stale — the count does not change, so nothing looks wrong. Observed 2026-08-14: the newest disclosure_date on the tape was still 2026-08-13, yet a row bearing 08-13 was first ingested at 07:01 ET the NEXT morning. A caller working from the previous afternoon's view of 08-13 missed $1.26M of buying in a position it already held. (5) SANITY-CHECK A BIG DOLLAR FIGURE AGAINST VOLUME, IN THIS API. total_value is derived (shares x price) wherever SEC did not file a total, so a filer's unit or decimal slip lands in it. The cheapest test is whether that many shares could plausibly have traded: get_daily_prices(ticker, since, until, include_ohlc=true) -> volume Compare the reported share count to the session's volume. A purchase that is a large multiple of everything that traded is worth a second look before acting on it. Verified examples, 2026-08-14: COE reported 592,320 ordinary shares against 52,418 ADS traded — a 60:1 ADS ratio, not a real $11.8M buy. EVGN reported 460,000 against 326,793 traded (141%) and was FINE — Evogene is dual-listed on NASDAQ and Tel Aviv, so the US tape sees only part of the volume. The check flags what to examine; it does not decide. Both readings beat ranking by total_value and trusting the top. RE-QUERY rather than reusing a prior window. 'Same disclosure date' does not mean 'same rows'. If you need to detect what is NEW since your last look, compare against the row identity you saw before rather than assuming a closed date is settled — and note that a sync timestamp moving is a statement about the JOB, not about the DATA. Pure-publisher posture: KeyVex mirrors SEC's exact bytes, documents these conventions and filer quirks rather than altering them, and never silently 'corrects' a value to KeyVex's guess of what was meant. A customer auditing KeyVex against EDGAR's source of record for each row (the bulk Form 345 extract for v2 rows) will find a byte-for-byte match. MACHINE-READABLE FLAGS — responses include a `source_metadata` block on rows where the above SEC-source patterns are detected. The block is keyed by field name, with an array of flag strings per field: `sec_perpetual_sentinel` (assertive — exact-string match on a known SEC sentinel value); `anomalous_year_likely_filer_entry` (calibrated — year outside the plausible range on transaction_date, exercise_date, expiration_date, or period_of_report; covers filing-pipeline data quality issues across the upstream-actor stack including filer typos, filing-agent default-epoch substitutions, and other cause-classes where the year falls outside any plausible range). Presence is the signal: clean rows have NO `source_metadata` field at all (not an empty object — the field is omitted entirely). Absence means 'no SEC source quirks detected,' NOT 'certified clean by audit' — agents weigh the difference. The raw source date values are preserved unchanged; the flag block carries KeyVex's labeled interpretation alongside, never replacing.

Returns 13F holdings — quarterly snapshots of equity positions held by institutional investment managers with $100M+ AUM, filed with the SEC. Each record is one (fund, security, quarter) tuple. Use this when the user asks about: which institutions hold a stock, a fund's portfolio, position changes quarter-over-quarter, or 'whale' activity in a specific name. Reporting lag: up to 45 days after quarter end. A 2026-Q1 filing typically appears in mid-May 2026. The most recent quarter visible always lags real time. Important: 13F covers institutional managers ≥ $100M AUM but does NOT include short positions, cash, options (with rare exceptions), or non-US-listed equities. It's a snapshot of long equity positions only. For 'did the fund increase its AAPL stake?' questions, check the position_change field — values are 'new', 'increased', 'decreased', 'closed', or 'unchanged' relative to the same fund's prior quarter. form: the filing a row comes from, '13F-HR' (the quarter's report) or '13F-HR/A' (an amendment). amendment_type: on an amendment, 'RESTATEMENT' (it replaces the report) or 'NEW HOLDINGS' (it adds positions); null on a 13F-HR. Both are absent on rows stored before they were recorded. Values are US dollars. 13F filings before 2023 report thousands and are multiplied by 1000; a filing that used the other unit is corrected once market prices confirm it: a 2023+ filing in thousands ×1000, a pre-2023 filing already in dollars ÷1000 (units_corrected true; value_as_filed holds the filed number, summed across a security's lines). A few filings whose units could not be confirmed are left as stored: a 2023+ one may read 1000x too low, a pre-2023 one 1000x too high. verification_status: VERIFIED means the filing matches its own totals (its rows add up to the row count and total value it declares). It is not a check against market prices: a filer's own consistent but wrong numbers still read VERIFIED, and so do a few filings stored before the thousands check began. INSUFFICIENT_DATA means that check failed or could not run, or the values look filed in the wrong unit (a 2023+ filing in thousands, or a pre-2023 filing in dollars) but could not be confirmed.

Returns the latest intraday price for ONE US-listed ticker, live from Tiingo's IEX feed; a stock's quote is reused for up to 10 seconds, so two calls in the same 10 s can return the same print. PAID PLANS ONLY. Use this when the question is 'what is X trading at now'. For price HISTORY (daily closes, splits, dividends) use get_daily_prices. Returns: price, and `as_of` — the exact timestamp of that quote, verbatim from the source. ALWAYS read `as_of` rather than assuming the quote is current: outside US market hours the feed returns the most recent session's final print, so a quote at 21:00 ET is a 16:00 ET price and `as_of` is how you can tell. `price_field` names the source field the price came from (tngoLast / last / mid) so a decision made on it can be re-derived later. A symbol the feed does not cover returns result: null with not_found_reason — never a substituted or stale price. Two reasons are possible: 'no_quote_for_ticker' (unknown symbol, or no quote available) and 'delisted_no_longer_trading' (the security stopped trading; the response carries delisted_since, the day after its last real trade). The second exists because the upstream feed keeps synthesising a current-session row for some delisted names — a flat zero-volume bar at the last real price, stamped with today's close — and a dead security has no current price to report. ⚠ VENUE COVERAGE IS IEX ONLY, and `price` is Tiingo's tngoLast derived from that feed — not a consolidated-tape print. This matters most on THINLY TRADED names: IEX is one venue carrying a few percent of US volume, so a symbol can go 20+ minutes mid-session without IEX seeing a trade. When that happens `as_of` legitimately reads stale DURING market hours. It means 'IEX has not seen an update', NOT 'the stock is not trading' — measured 2026-09-09, BWFG sat at a 22-minute-old as_of while trading normally. Do not treat a stale as_of on a thin name as a fault. Three fields let you judge that for yourself rather than trusting a verdict this tool does not make. `iex_session_volume` is the session volume ON IEX ONLY — it is NOT consolidated volume and is typically a small fraction of it, so never compare it against share counts from other endpoints. `prev_close` is the prior session's close. And `flat_session` is true when all four OHLC legs equal the price AND volume is zero AND the price equals prev_close — the shape the feed manufactures for a security that no longer trades. It is a SHAPE, not a delisting verdict: stable-NAV money market funds hold a constant price by design and look identical. It is NULL when the test was impossible — a missing OHLC leg or an unreported volume — and null means 'could not test', never 'false'. One call = one ticker, by licence — there is no multi-ticker parameter. Close Prices from Tiingo.com.

Returns SEC Form ADV registry records — every SEC-registered investment adviser (~17K RIAs) and exempt reporting adviser (~6.5K ERAs, mostly private-fund advisers), from the SEC's monthly roster extract. One record per firm (CRD number) with regulatory AUM (discretionary / non-discretionary / total, Item 5F), employees and IA reps, client counts, custody flags (Item 9A), and disciplinary disclosure flags (Item 11, verbatim sub-question codes). Use this when the user asks: who advises/manages money, how big is an adviser, largest RIAs by state, advisers with disciplinary history, or to vet a firm before pairing with enforcement / holdings data. firm_type: 'registered' RIAs report regulatory AUM; 'exempt_reporting' ERAs do NOT report Item 5F — their AUM fields are null by construction (they report private-fund data instead; see adviserinfo_url for Section 7.B detail). Registry posture: this is a CURRENT-ROSTER snapshot refreshed monthly, not an event history. snapshot_month is the last month the firm appeared — a stale snapshot_month means the firm dropped off the roster (deregistered). min_aum filters on total regulatory AUM and requires the default aum_total sort. CIK is the join key to EDGAR datasets (13F institutional holdings, enforcement). v1A maps a curated ~30-field subset of Form ADV Part 1A's 448-column grid; adviserinfo_url links the firm's full IAPD page. Pure-publisher posture: the SEC's roster as published. A disciplinary FLAG is a disclosure, not a verdict — agents read the detail on IAPD.

Returns Lobbying Disclosure Act (LDA) filings — quarterly LD-2 reports filed by registered lobbyist firms with the Senate Office of Public Records. Each record covers one (registrant, client, quarter) tuple, listing income paid, issues lobbied on, and government entities contacted. Use this when the user asks about: who's paying lobbyists, what issues a company is lobbying on, which senators or agencies a firm is contacting, lobbying spend by industry or sector, or to cross lobbying activity against congressional trades or federal contracts for political-influence analysis. Each filing has a `lobbying_activities` array (one entry per issue area worked on) plus three flattened summary arrays at top level: - general_issue_codes: 3-char codes (DEF, HEA, TRA, ENV, FIN, ...) - government_entities: agencies/branches contacted - lobbyist_names: lobbyists who worked the issue Top-level arrays support indexed queries; the nested array carries issue-level descriptions and lobbyist position info. `general_issue_codes` filter is OR-semantic — pass an array, match any filing containing AT LEAST ONE of those codes (max 30, per Firestore array-contains-any). Examples: ['DEF'] for defense, ['HEA','MMM'] for health + Medicare/Medicaid, ['TAX','FIN'] for tax + financial services. Income vs expenses (IMPORTANT for ranking by spend): the LDA mandates a hard split. Third-party lobbying firms report `income` (what the client paid them); in-house corporate lobbying departments report `expenses` (what they spent). The two fields are mutually exclusive — any given filing has one or the other, not both. In practice ~30% of filings have income, ~70% have expenses, with a small population reporting neither (administrative registrations). So `sort_by=income` ranks the third-party-firm subset; for an actual top-spenders leaderboard, agents should fetch both populations and sum `income + expenses` per filing client-side. v1.1 polish will add a derived `total_lobbying_spend` field that does this sum server-side for indexed queries. client_is_government is true when the client is a government body (US states often hire lobbyists). Activity descriptions are truncated at 5000 chars during ingestion to stay under Firestore's per-doc cap; agents can fetch the full filing via filing_document_url for the unbounded prose.

Returns LD-203 semiannual contribution reports — what registered lobbyists and lobbying firms themselves contribute: FECA campaign contributions, honorary expenses, event/meeting costs, and presidential-library / inaugural-committee donations, each item naming the HONOREE (the covered official who benefited). This is the reverse angle of get_lobbying_filings: filings show who pays lobbyists; LD-203 shows where the lobbyists' own money goes. Coverage: 2008→present (~40K filings/year; roughly half are 'no contributions' certifications, excluded by default — set include_empty=true to see them). Record shape: one record per filing — filer (lobbyist name or registrant firm), filing_year + period (mid_year | year_end), nested contribution_items[] (contribution_type, contributor_name, payee_name, honoree_name, amount, date), flattened honoree_names[] / payee_names[] / contribution_types[], and contributions_total_usd (simple sum of item amounts). Filters: honoree_name is the political join — substring against any item's honoree (e.g., 'schumer'). registrant_name matches the firm; lobbyist_name the individual filer; payee_name the receiving committee. contribution_type exact values: 'feca' (campaign money), 'honorary', 'meeting', 'presidential_library', 'inaugural_committee'. Cross-source: pair with get_fec_contributions (the FEC's view of the same FECA money, itemized ≥$200), get_lobbying_filings (the same registrant's client work), get_member_profile (resolve the honoree to party/state/committees). Pure-publisher posture: filings as posted to the Senate LDA system; contributions_total_usd is arithmetic, not a score.

Returns Form 8-K filings — the SEC's 'current report' form, filed within 4 business days of any material event at a publicly-traded company. Each record is one filing, with `item_codes` declaring WHAT kind of event(s) it covers. Use this when the user asks about: recent CEO/CFO departures or appointments, M&A announcements, earnings releases, big contract wins, restructurings, going-concern warnings, exec compensation changes, or any 'what just happened at this company' question. Item codes (most-used; many more exist): 1.01 Entry into a Material Definitive Agreement 1.02 Termination of a Material Definitive Agreement 2.01 Completion of Acquisition or Disposition of Assets 2.02 Results of Operations (earnings releases live here) 2.03 Creation of a Material Direct Financial Obligation 3.01 Notice of Delisting / Failure to Satisfy Listing Rule 3.02 Unregistered Sales of Equity Securities 4.01 Changes in Registrant's Certifying Accountant 5.02 Departure / Election / Appointment of Officers + Directors 5.07 Submission of Matters to a Vote of Security Holders 7.01 Regulation FD Disclosure 8.01 Other Events (catch-all) 9.01 Financial Statements and Exhibits — NOTE: nearly every 8-K ticks this 'paperwork box.' Searching JUST for 9.01 returns the firehose; combine it with another item_code to focus. `item_codes` filter is OR-semantic: pass an array, match any filing containing AT LEAST ONE of those codes. Capped at 30 codes per query (Firestore array-contains-any limit). Examples: ['5.02'] for exec changes; ['1.01','2.01'] for any deal activity (LOI or close); ['2.02'] for earnings. Amendments (8-K/A) get their own row with `is_amendment: true`. The original 8-K stays in place. v1 does NOT populate `original_accession_number`; agents can find candidates by matching (ticker, period_of_report) across rows. Filter `is_amendment: false` for clean original-only views. v1 does not extract the prose body — `primary_document_url` points agents at the source HTML for direct fetch. The structured items are what's queryable here.

Returns Congressional member profiles from the unitedstates/ congress-legislators catalog. Each record is one current House Representative or Senator, keyed by bioguide_id (the permanent member identifier — e.g., 'C001035' for Susan Collins). Use this when the user asks about: which committees a member sits on, who chairs the Senate Banking Committee, all Republicans on House Armed Services, party/state/district lookup for a specific member, or to enrich congressional_trades records with member context (party + state + committee assignments). Filter by bioguide_id for a direct fetch; by member_name for a case-insensitive substring search; by committee_id (e.g., 'HSAS' for House Armed Services, 'SSAF' for Senate Agriculture, 'HSAG15' for the Forestry & Horticulture subcommittee under House Ag) to find all members of a committee. Combine state + chamber + party for caucus-level queries. Committee codes follow the Library of Congress 'Thomas' convention: House full committees: HSAG (Agriculture), HSAS (Armed Services), HSAP (Appropriations), HSBA (Financial Services), HSED (Education), HSEN (Energy & Commerce), HSII (Natural Resources), HSJU (Judiciary), HSWM (Ways and Means), etc. Senate full committees: SSAF (Ag), SSAS (Armed Services), SSAP (Appropriations), SSBK (Banking), SSCM (Commerce), SSEG (Energy), SSFI (Finance), SSHR (HELP), SSJU (Judiciary), etc. Subcommittees append the subcommittee thomas_id: HSAG15, HSBA00. Photo URLs are constructed (theunitedstates.io/images/congress/ original/{bioguide_id}.jpg) but Cloudflare-protected — clients fetch directly. Senate class field (1/2/3) on senators only.

Returns Form N-MFP3 monthly money-market fund reports — one record per (fund series, month): fund category (Government / Prime / Single State…), net assets, shares outstanding, weighted average maturity (wam_days) and life (wal_days), the fund's DAILY daily/weekly liquid-asset percentages for the month (verbatim fractions of 1 — the money-market stress series), monthly gross subscriptions/redemptions ON N-MFP3 ONLY, and stable-NAV posture. ⚠ MONTHLY FLOWS ARE NOT ON EVERY RECORD. gross_subscriptions_month / gross_redemptions_month are filed per SHARE CLASS on N-MFP3 and are served as the sum across a filing's classes. N-MFP2 and N-MFP do not ask for monthly flows at all (N-MFP2 reports weekly), so those rows carry null — the source's silence, not ours. Read `flows_basis` to tell them apart: "monthly_sum_of_classes" or "not_filed_monthly". `form_type` says which form the record came from. ⚠ A SUM COVERS ONLY THE CLASSES THAT REPORTED. flows_classes_reporting_subscriptions / _redemptions say how many did; compare each against total_share_classes, which is how many EXIST. The two counts are separate because a class can report one figure and omit the other. WHERE THOSE FIELDS ARE ABSENT, COVERAGE IS NOT RECORDED — that is NOT a statement that the sum is complete. Rows written before 2026-09-11 predate the fields and are not backfilled for them. Use this when the user asks about: money-market fund assets or flows, fund liquidity levels, WAM positioning as a rates signal, prime-vs- government fund dynamics, or a specific fund family's money funds. The adviser_file_number (801-…) joins get_investment_advisers for the manager's full ADV profile; registrant cik joins other EDGAR datasets. Each fund files monthly — filter series_id + sort report_date asc to read one fund's history; filter by report_date (since/until) for a cross-fund month snapshot. Coverage: 2010-11→present across all THREE form generations, with per-era cadence differences kept verbatim: N-MFP3 records (2024-06→) carry DAILY liquidity/shadow-NAV/yield series with real dates; N-MFP2 records (2016-10→2024-06) carry WEEKLY Friday points labeled with the source's own fridayWeek1..5 keys in the date field (the filing reports week numbers, not dates — never fabricated); original N-MFP records (2010-11→2016-10) carry single month-end shadow-NAV + yield points and NO liquidity percentages (that reporting began with the 2014 reforms). series_name is empty before 2024 (not in the older XML). The per-security portfolio schedule and per-class yields live in the filing XML — follow source_url (v1.1 scope). Pure-publisher posture: the fund's reported numbers as filed, parsed by KeyVex — no derived stress scores; liquidity thresholds are for agents to apply.

Returns NLRB (National Labor Relations Board) case filings: unfair- labor-practice charges and union representation/election petitions. Use this when the user asks about: union organizing at a company, labor disputes or ULP charges, union election petitions and outcomes, decertification efforts, or a company's labor-relations record. Case numbers follow {region}-{type}-{sequence}, e.g. '03-CA-390171' (NLRB Region 03, CA charge). The middle code determines the case family: C-cases are ULP CHARGES against an employer (CA) or a union (CB, CC, CD, CE, CG, CP) — allegations of unlawful labor practices. R-cases are REPRESENTATION petitions — RC (union seeks certification), RD (employees seek decertification), RM (employer-filed), plus UD/UC/AC unit matters. Each record carries case_type ('ULP' or 'representation') and case_subtype (the raw code). The `name` field is the named party on the filing — usually the employer, but on CB/CC-type charges it is the union being charged. employer_name matches against it as a substring. Representation cases carry eligible_voters, certified_representative (filled after a won election), and unit_sought (the bargaining-unit description). FRESHNESS CAVEAT: cases mutate after filing — status flips Open→Closed and date_closed / reason_closed / certified_representative fill in later. The daily sync re-pulls a trailing 180-day date_filed window, so those fields on cases FILED MORE THAN ~180 DAYS AGO may lag the source until a periodic full re-pull; the source_url case page is always current. Filter combinations note: server-side indexes support ONE of state / case_type combined with the date_filed sort + since/until. Other filters (employer_name substring, case_subtype, status, region, state+case_type together) post-filter client-side over a widened fetch window. Pure-publisher posture: NLRB's public case rows as published — no outcome scoring. Each record's source_url links to the nlrb.gov public case page (which also carries docket activity and related documents not in this dataset).

Returns IRS Form 990-series e-filing records — the registry of every e-filed nonprofit return the IRS has released, 2017→present (~5.5M filings; ~400-750K/yr). One record per return: EIN, organization name, return type (990 = full; 990EZ = small; 990PF = private foundation; 990T = unrelated business income; the 2019-era index also carries IRS codes 990EO/990O verbatim), the tax period covered (YYYYMM), and the IRS release year. Use this when the user asks: does nonprofit X file with the IRS, when did a foundation last file, which returns has an EIN filed, or to anchor a nonprofit's identity (EIN) before joining grants / lobbying / OIG data by name. Records carry the filing's extracted FINANCIALS: total_revenue, total_expenses, total_assets_eoy, net_assets_eoy, and officers[] — top 25 by reported compensation with name/title. Coverage (reconciled 2026-07-08): 100% for release years 2019-2026, 97% for 2018, 69% for 2017 — the shortfall is IRS-side (the pre-2017 XML archives that held those filings' documents were retired by the IRS; the registry rows remain, without financials). Meanings follow the form: for 990T, total_revenue is unrelated-business taxable income. Officer compensation is as reported to the IRS. sub_date is set only for 2019-era records (later IRS indexes carry only the year — KeyVex never fabricates dates); tax_period is the reliable time axis. A nonprofit's fiscal year varies — tax_period 202506 means the period ENDING June 2025. Pure-publisher posture: the IRS index rows as published.

Returns SEC Form N-PORT filings — monthly portfolio reports from registered investment companies (mutual funds, ETFs, closed-end funds). Use this when the user asks about: recent fund portfolio filings, when a specific fund family last reported, monthly cadence of fund disclosures, or to bridge from a fund trust name to the primary_doc.xml that contains full per-holding portfolio detail. Source: SEC EDGAR full-text search. Covers both NPORT-P (original filing) and NPORT-P/A (amendments). N-PORT is filed within 60 days of each month-end; period_ending tells you which month the report covers. v1A returns metadata only: filer trust name + CIK, period_ending, filing type, SEC investment company file number (e.g., '811-21864'), filer state + state of incorporation, and the URL to the full primary_doc.xml. Per-holding portfolio detail (every security in the fund's portfolio with quantity, fair value, currency, etc.) lives in that XML — agents follow the URL when they need security-level data. Pairs with get_institutional_holdings (13F): 13F is quarterly, filed by INVESTMENT MANAGERS (Berkshire, Vanguard, BlackRock); N-PORT is monthly, filed by the FUND TRUST. Together = fresher snapshots across two complementary universes (manager-level vs fund-level).

Returns OFAC Specially Designated Nationals (SDN) sanctions list entries, republished as-is (not a screening service; verify against treasury.gov). Use this for: sanctions-program queries (e.g., 'who's on the Russia SDN list'), or cross-referencing named individuals / entities against the canonical US sanctions list. Source: US Treasury OFAC — sanctionslistservice.ofac.treas.gov. ~19,000 entries refreshed daily. Each entry represents a person, entity, vessel, or aircraft sanctioned by the US government under one or more programs (CUBA, IRAN, SDGT [terrorism], NPWMD [WMD proliferation], RUSSIA-EO14024, etc.). US persons (citizens, residents, US-domiciled companies) are legally prohibited from transacting with SDNs — this is the canonical list published by OFAC. Filter by name substring for primary lookups. entity_type values: 'individual', 'entity', 'vessel', 'aircraft'. Every SDN record carries exactly one of the four — companies are 'entity'. program is a substring filter against the comma-delimited Program field (e.g., 'iran', 'russia', 'narcotics'). remarks substring catches aliases, DOB / passport references, and related-party hints. Direct ent_num lookup is fastest (OFAC's stable entity number). WHAT'S NOT IN v1A (data-model limitations to know about): the schema does NOT include designation_date (when OFAC originally added the entry). OFAC's basic SDN.csv source file only provides 12 columns and omits this — the date lives in OFAC's advanced XML and a separate 'Recent Actions' page on their site. So 'sanctions added in the last N days' is not directly queryable via this tool — point users at ofac.treasury.gov/recent-actions for that specific question. v1.1 polish will add advanced-XML ingestion to capture designation_date. Also: there's no since/until filter and no date sort option for the same reason — the only sort options are name and ent_num. Pure-publisher posture: KeyVex returns OFAC's published list as-is. No derived 'risk score' or 'similarity match' — agents handle fuzzy matching downstream. For broader list coverage, agents should also consult the US Consolidated Screening List (get_screening_list) which spans 12 export-control / sanctions lists from State + Commerce + Treasury.

Returns entries on the HHS Office of Inspector General 'List of Excluded Individuals/Entities' (LEIE). Anyone on this list is barred from billing Medicare, Medicaid, or any federal healthcare program. Updated monthly by OIG; KeyVex re-scrapes monthly and overwrites. Use this when the user asks about: healthcare-fraud exclusions, Medicare/Medicaid program-integrity research (not employment or eligibility decisions about individuals — Terms §8A), geographic concentration of exclusions, or a specific person/business listed on LEIE. Cross-source tip: pair with get_federal_contracts to flag contractors who appear on the exclusion list. A government contractor with an OIG exclusion is worth checking against the official LEIE at oig.hhs.gov. Statutory exclusion types (the most common): - 1128a1 Conviction of program-related crimes - 1128a2 Conviction relating to patient abuse - 1128a3 Felony conviction relating to healthcare fraud - 1128a4 Felony conviction relating to controlled substances - 1128b4 License revocation, suspension, surrender - 1128b5 Exclusion or suspension under federal/state healthcare - 1128b7 Fraud, kickbacks, and other prohibited activities - 1128b8 Entities controlled by a sanctioned individual Scope: the LEIE lists only CURRENTLY-ACTIVE exclusions — OIG removes a party once reinstated (reinstatements are a separate OIG publication not ingested here). So every record is, by definition, an active exclusion, and reinstatement_date is effectively always empty. Pure-publisher posture: we surface the listing as-published. Some names match common-name individuals who aren't the excluded party — the agent / user is responsible for context disambiguation (DOB, address, NPI).

Returns CMS Open Payments records — the Sunshine Act database of every payment / transfer of value from drug + device manufacturers and GPOs to US physicians, non-physician practitioners, and teaching hospitals (~15M records per program year, 2019→present). LIVE passthrough to CMS's own API: results reflect CMS's current data and `total_count` is CMS's authoritative count for the filtered query (the results array is just the requested page). Use this when the user asks about: pharma/device money to doctors, a company's physician-payment footprint, speaker-fee / consulting / royalty programs, industry funding of research (with ClinicalTrials.gov IDs), or physician ownership stakes in manufacturers. payment_type selects the dataset (schemas differ; rows are CMS's fields verbatim): general (default) — meals, travel, consulting, speaker fees, royalties, honoraria. Fields incl. nature_of_payment_or_transfer _of_value, name_of_drug_or_biological_or_device_or_medical _supply_1, covered_recipient_specialty_1. research — research payments incl. name_of_study, clinicaltrials_gov_identifier, preclinical_research_indicator. ownership — physician ownership/investment interests (total_amount_invested_usdollars, value_of_interest, terms_of_interest); recipient fields are physician_*. summarize=true returns CMS's own pre-aggregated per-(company, nature) totals for the year — transaction counts + dollar totals per payment nature. The nature codes in that dataset ship without a public CMS legend; KeyVex labels the seven codes it has VERIFIED by exact count+total reconciliation against detail data (1=Consulting Fee, 2=Speaker/faculty compensation, 6=Food and Beverage, 7=Travel and Lodging, 9=Charitable Contribution, 10=Royalty or License, 14=Grant); unverified codes pass through with an empty label rather than a guess. Matching: company is a substring (matches subsidiaries: 'pfizer' catches PFIZER INC.); recipient names are EXACT (CMS stores uppercase; we uppercase for you); npi is the exact National Provider Identifier — the precise join key to get_oig_exclusions. Payments are attributed to the manufacturer AS FILED — no ticker/CIK; try the operating-company name. Program years: 2019 through the latest published year (CMS refreshes semiannually; year defaults to the latest). Pagination: limit ≤ 500 per page (CMS cap), use offset for more. Pure-publisher posture: CMS's records as filed, parsed by KeyVex (the source record is authoritative) — no derived influence scores. Disclosure ≠ wrongdoing; these are lawful, statutorily-disclosed payments.

Returns OSHA workplace-safety enforcement records from the Department of Labor's enforcement data: inspection cases (who was inspected, where, why, when) with optional violation citations attached (standard cited, violation type, penalties, abatement dates). Use this when the user asks about: a company's workplace-safety record, OSHA penalties or citations, fatality/catastrophe investigations, inspection activity by state or industry (NAICS), or contractor safety research. Result rows are INSPECTIONS. Pass include_violations=true to attach each inspection's citations under a `violations` array (or pass activity_nr for a direct lookup, which always includes them). min_penalty keeps only inspections with at least one violation whose initial or current penalty meets the threshold (implies include_violations). insp_type codes (DOL's own legend): A=Accident, B=Complaint, C=Referral, D=Monitoring, E=Variance, F=FollowUp, G=Unprog Rel, H=Planned, I=Prog Related, J=Unprog Other, K=Prog Other, L=Other-L, M=Fat/Cat (fatality/catastrophe), N=Unprog Emph. Each record carries insp_type_label with the decoded value. Violation viol_type codes: S=Serious, W=Willful, R=Repeat, O=Other, U=Unclassified. Violations carry delete_flag='X' when the source later deleted the citation — rows are kept with the flag, never dropped. Filter combinations note: server-side indexes support ONE of state / naics_code combined with the open_date sort + since/until. Other filters (establishment_name substring, insp_type, state+naics together, min_penalty) post-filter client-side over a widened fetch window. sort_by=case_mod_date is the 'recently updated cases' firehose and cannot be combined with state / naics_code / since / until. Pure-publisher posture: DOL's enforcement rows as published — codes kept verbatim (with the source's own legend decoded alongside), no safety scoring. Each record's source_url links to the osha.gov establishment inspection-detail page.

A ticker search also returns rows this issuer FILED UNDER SYMBOLS IT NO LONGER LISTS (renames, filer typos, ADR spellings). Each row keeps `ticker` as SEC received it and gains `current_ticker` when the issuer trades under a different symbol today. Separately-listed share classes are NOT merged, and asking for a RETIRED symbol returns only rows filed under it — retired symbols get reissued to other companies. Returns Form 144 filings — notices of proposed sale by corporate insiders (officers, directors, 10%+ holders) under Rule 144 of the Securities Act. Each record is one planned-sale line from one filing. ⚠ aggregate_market_value is NULLABLE. A Form 144 that did not state a value now reports null rather than 0 — but a filer who genuinely stated 0.00 still reports 0, and that happens. Null never satisfies min_value and sorts last. ⚠ AND SOME FILERS STATE THE ISSUER'S MARKET CAP IN THAT BOX, WHICH PUTS THEM AT THE TOP OF A DESCENDING VALUE SORT. Measured 2026-09-04: 5 of the top 100 — SYF 4,000 shares stating $25.24bn ($6.31m per share), IT 860 shares stating $11.72bn. Dividing those by shares_outstanding on the same row gives $77.57 and $185.66, which are the real share prices. ⚠ THE CHECK CATCHES TWO OF THOSE FIVE, NOT ALL FIVE, AND SAYS SO RATHER THAN OVERSTATING ITSELF. TCMD, EPSM and XHR imply $36,017, $17,577 and $16,686 per share — absurd for those issuers, but below BRK.A's real $740,000 peak, so no per-row test separates them from a genuine high-priced sale. They still read 'checked'. Settling them needs a comparison against the day's actual price bar. Every row therefore carries aggregate_market_value_check: 'checked' the value implies a plausible price for the shares sold 'not_stated' no value was filed (distinct from a filed 0.00) 'no_share_count' no share count, so the check could not run 'implausible_looks_like_market_cap' implies a per-share price above any that has traded, and shares_outstanding yields a plausible one instead 'implausible_unexplained' implies an impossible price and shares_outstanding does not explain it — we can say it is not the sale value without being able to say what it is ⚠ The number is NOT corrected. SEC's bytes are served exactly as filed, and we do not invent a value the filer never stated. If you rank by this field, exclude anything whose check does not read 'checked' — otherwise the top of your list is market capitalisations. Use this when the user asks about: insiders who have announced they're about to sell, upcoming insider sales at a specific company, large planned sales by value, or which executives are signaling intent to exit positions. Form 144 is a *forward-looking* signal. It's filed BEFORE the actual sale, which later lands as a Form 4. The complement to get_insider_transactions: that tool tells you what insiders just did, this one tells you what they're about to do. Filing thresholds: ≥5,000 shares OR ≥$50,000 aggregate value. The aggregate_market_value is the insider's estimate at filing time; the actual sale price/value can differ. The approximate_sale_date is also an estimate — the real Form 4 transaction_date may be days later. Most Form 144 filings list one security line, but a single filing can cover multiple share classes (e.g., separate Class A + Class B). Each line is returned as its own record.

Returns SEC Form D filings — Reg D / Rule 506 private placement offering notices. Use this when the user asks about: who's raising private capital right now, new VC fund formations, private equity raises, real-estate syndicates, hedge fund launches, who's claiming Rule 506(b) vs 506(c) exemption, or to identify directors / executive officers of newly-formed entities. ⚠ total_amount_sold, min_investment_accepted, total_number_already_invested, sales_commissions and finder_fees are NULLABLE — a Form D that did not state a figure reports null rather than 0. The distinction matters here more than anywhere: a Form D filed at the START of an offering legitimately reports $0 sold, so 0 and null mean genuinely different things. Rows stored before 2026-08-17 cannot tell you which they were. Null never satisfies min_amount_sold and sorts last. Source: SEC EDGAR full-text search + per-filing primary_doc.xml. All Reg D filings (504 / 506(b) / 506(c)) plus Section 4(a) exempt offerings flow through here. Form D must be filed within 15 days of the first sale. Each record carries: issuer entity (name, CIK, address, jurisdiction of incorporation, entity type), offering data (industry group, investment fund type for pooled funds, total offering / sold / remaining, minimum investment, federal_exemptions claimed), filing metadata (file_date, date_of_first_sale, is_amendment), and a related_persons[] array of directors / executive officers / promoters. Federal exemption codes (federal_exemptions array): 06b — Rule 506(b) (no general solicitation; up to 35 non-accredited) 06c — Rule 506(c) (general solicitation OK; all accredited) 04(2) — Section 4(a)(2) (statutory private placement) 3C — ICA Section 3(c) (3(c)(1), 3(c)(5), 3(c)(7) etc. — common for funds) 3C.1 — ICA 3(c)(1) (up to 100 investors) 3C.7 — ICA 3(c)(7) (qualified purchasers only) Common industry_group_type values: 'Pooled Investment Fund' (with investment_fund_type='Venture Capital Fund' | 'Private Equity Fund' | 'Hedge Fund' | 'Other Investment Fund') 'Technology', 'Real Estate', 'Health Care', 'Energy', 'Financial Services', 'Manufacturing', 'Other'. Direct filing_id lookup is fastest (accession number). Substring filters on issuer_name, industry_group_type, investment_fund_type, and jurisdiction_of_inc enable topic-style queries. Combine federal_exemption + min_amount_sold for 'who's raising real money under 506(c)' analyses.

Returns safety recalls from federal agencies — drug recalls (FDA), medical device recalls (FDA), food/dietary supplement recalls (FDA), consumer-product recalls (CPSC), and vehicle, equipment, tire and child-restraint recall campaigns (NHTSA, since 1966). Use this when the user asks about: recent recalls for a specific company or product, FDA Class I (most severe) recalls, food contamination recalls, drug shortages and recalls, vehicle recalls by make or model (e.g. all Ford F-150 recalls), or to add a 'product-safety event' flag to insider activity / 8-K filings / enforcement actions. Sources (filter via the `source` enum): fda_drug — openFDA /drug/enforcement.json. Drug recalls including prescription, OTC, biologics. Class I/II/III severity. fda_device — openFDA /device/enforcement.json. Medical device recalls (implants, diagnostics, equipment, software). Same classification scheme. fda_food — openFDA /food/enforcement.json. Food + dietary supplements. Pathogen contamination, allergen mislabeling, etc. cpsc — saferproducts.gov RestWebServices/Recall. Consumer-product recalls (clothing, electronics, toys, batteries, etc.). No severity classification; classification field is null. nhtsa — NHTSA recall campaigns from NHTSA's flat files, updated daily. One row per campaign (recall_number = NHTSA's campaign number, e.g. '26V652000'); recalling_firm = the manufacturer that filed the defect report; product_description = NHTSA's defect summary; reason_for_recall = its consequence; remedy and notes as NHTSA writes them. Consumer advisories first: do_not_drive (NHTSA advises owners not to drive the vehicle until it is repaired) and park_outside (park outside, away from structures: fire risk); null on campaigns that predate the advisories. product_quantity is NHTSA's potential units affected (null when not stated); vehicle_model_year_count is the number of distinct make/model/year combinations, not units. makes lists up to 20 makes ('+N more' beyond). A campaign's full vehicle list comes only with source=nhtsa + recall_number: `vehicles` (500 per page, by make / model / year), `vehicles_total`, and `vehicles_next_cursor` (pass it back as vehicles_cursor). classification and status are null (NHTSA uses neither). A campaign NHTSA's file stops listing keeps its row: nhtsa_listing says 'no longer listed by NHTSA' and nhtsa_unlisted_since is the date KeyVex first observed it missing (NHTSA publishes no removal date or reason); both are null while NHTSA lists it. Source: NHTSA Office of Defects Investigation; public domain (http://www.usa.gov/publicdomain/label/1.0/). Cross-source pairing pattern: Recall → 8-K Item 7.01/8.01: pair with get_material_events Recall → insider sells: pair with get_insider_transactions Recall → SEC/DOJ follow-on: pair with get_enforcement_actions Recall → company filings: pair with get_proxy_filings (DEF 14A risk factors) Each record is one recall. Identifier format: `{source}-{recall_number}` (e.g., 'fda_drug-D-1234-2026'). FDA classifications: Class I — serious adverse health consequence or death Class II — temporary or reversible health consequence Class III — unlikely to cause adverse health consequence Source freshness (per-source publication cadence, not KeyVex bug): CPSC publishes within ~1-2 days; recent data flows hourly-fresh. openFDA's snapshot updates every ~10-14 days, and each snapshot carries recall_initiation_date values that LAG the snapshot date by another 30-45 days (the time between FDA classifying a recall and openFDA exposing it). Net: FDA records in this collection typically run ~4-6 weeks behind real-world recall dates, while CPSC is current. A default desc-by-date sort therefore looks CPSC-heavy at the top even when FDA matters more for the query. Filter by source='fda_*' to see FDA-only and avoid the skew. Searching recalling_firm / product_description: case-insensitive; 4+ characters match anywhere, 1-3 characters match a whole word ('3M', 'GE'). A term of 3+ characters searches EVERY recall; a 1-2 character term searches only the newest 5,000 recalls the query's other filters select (the oldest, with sort_order=asc), and coverage_warning says so. Vehicle filters (vehicle_make, vehicle_model) match NHTSA campaigns only, through their vehicles: vehicle_make is the exact make, case- and space-insensitive ('ford' = 'FORD'); vehicle_model follows the search rule above (3+ characters search every campaign; 1-2 characters the newest 5,000). classification filter scope: 'classification' is an FDA-only field. CPSC and NHTSA records always have classification=null (neither uses the FDA severity scheme). Filtering by classification excludes ALL CPSC and NHTSA rows by definition. The query response surfaces this with a notice in coverage_warning when the filter is set.

Returns Schedule 14A proxy filings — the document public companies send shareholders ahead of annual or special meetings. Each record is one filing carrying executive compensation tables, board nominations, shareholder proposals, auditor info, and voting matters. Use this when the user asks about: executive compensation, board elections, shareholder proposals, M&A votes, proxy contests, auditor changes, say-on-pay outcomes, or upcoming annual meetings. Coverage: the full DEF 14A family back to 2016 for the US public-company universe (sourced from EDGAR's complete quarterly full-index); a daily feed keeps it current. Rows are tagged with a company's PRIMARY common ticker — for dual-class issuers (e.g. GOOGL/GOOG, BRK-A/BRK-B) query by company_cik to retrieve every share class in one shot. Filing types (the four-form DEF 14A family): DEF 14A — Definitive proxy (the annual-meeting filing) DEFA14A — Additional materials (supplements to a prior DEF 14A) DEFM14A — Merger-related proxy (filed when shareholders vote on M&A) DEFR14A — Revised definitive proxy (amendments to a prior DEF 14A) Convenience flags derived from filing_type: is_merger_related — true for DEFM14A is_amendment — true for DEFR14A is_additional_materials — true for DEFA14A period_of_report population (IMPORTANT for filtering / sorting): - Recent-window rows (filed ~2024-onward via the daily feed / per-ticker pull): DEF 14A primaries ~100% populated (meeting/record date); DEFA14A/DEFM14A/DEFR14A typically EMPTY (correct-as-filed — SEC's submissions API leaves those reportDate fields blank). - Historical backfilled rows (the bulk of 2016-2024 depth, sourced from EDGAR's full-index): period_of_report is EMPTY for all form types — the index carries no report date. - Bottom line: filter/sort by filing_date for chronological queries; period_of_report is not reliably present across the collection. v1A is metadata-only: ticker, company name, CIK, filing type, dates, primary document URL. The proxy body is not extracted in v1. `primary_document_url` points agents at the source HTML for direct fetch when they need exec comp tables or proposal text.

Returns SEC Form 1-A filings — Regulation A+ 'mini-IPO' offering statements, 2015-06→present: companies raising up to $20M (Tier 1) or $75M (Tier 2) from the public without a full IPO. One record per filing with the issuer (SIC code, jurisdiction, year incorporated, employees, city/state), tier election, offering terms (security types, count, price, total aggregate amount, estimated net), service providers WITH FEES (underwriter, sales commissions, auditor, legal), and the issuer's summary financials from Part I (cash, assets, liabilities, equity, revenues, net income). Use this when the user asks about: Reg A / Reg A+ raises, mini-IPOs, small- cap capital formation, who's underwriting or auditing small offerings, or issuer financials before a raise. form family: '1-A' initial | '1-A/A' amendment | '1-A POS' post-qualification amendment (is_post_qualification) | -W withdrawals (is_withdrawal, metadata-level). One offering typically chains 1-A → 1-A/A… → qualification → 1-A POS updates — filter cik + sort asc to read it. Coverage (honest): the offering-statement family only. The Reg A+ periodic reports — 1-K annual (with actual proceeds raised), 1-SA semiannual, 1-Z exit — use different schemas and are a planned separate dataset. Offering circulars (253G) are prose documents — follow filing_index_url. Pure-publisher posture: issuer-reported Part I numbers as filed, parsed by KeyVex.

Returns SEC Form S-1 / S-3 / S-3ASR registration statements — securities offering registrations filed with the SEC. Use this when the user asks about: which companies are going public (IPO pipeline via S-1), shelf registrations (S-3 / S-3ASR — company registers securities to sell over multiple offerings without re-registering; large established issuers use the automatic S-3ASR variant), recent secondary offerings, registration amendments updating prior filings, or to bridge from a company name / ticker to the prospectus prose. Forms covered: S-1 — Initial registration (IPO + first-time registrants) S-1/A — Amendment to an S-1 S-3 — Shelf registration (issuers meeting reporting / market-cap criteria; lets them issue securities over time without re-registering each time) S-3/A — Amendment to an S-3 S-3ASR — Automatic shelf registration. The shelf form used by Well-Known Seasoned Issuers (large established companies like Apple, Ford, most of the S&P 500). Effective on filing. These issuers file S-3ASR, NOT plain S-3. Source: SEC EDGAR full-text search. Returns one record per filing, deduped by accession. Exhibit attachments (EX-10, opinion letters, fee tables, etc.) are filtered out — only the canonical form types are returned. v1A is metadata only. Each record has filer name + CIK + optional ticker + SEC file_number, state, SIC code(s), and URLs. Substantive prospectus content (offering size, share counts, use of proceeds, risk factors, financial statements) lives at primary_document_url — agents follow for the prose. SCOPE — covers S-1 (IPO), S-3 + S-3ASR (shelf, including WKSI auto shelves), plus /A amendments. S-8 employee-benefit-plan registrations, S-4 merger/acquisition registrations, and F-series foreign-issuer forms are NOT ingested. 424B prospectus supplements (offering takedowns off an existing shelf) are out of scope — query the shelf registration itself. Amendment chains: all amendments share the same sec_file_number as the original. Use sec_file_number filter to fetch an entire amendment chain. Pure-publisher posture: KeyVex doesn't derive 'likely-to-IPO' or 'price-target' signals from registration filings.

Returns congressional roll-call vote metadata (House + Senate) from api.congress.gov. Use this when the user asks about: recent votes in either chamber, votes on a specific bill, votes by date range, or to chain to per-member positions via the source_data_url. Sources: api.congress.gov v3 for House votes; senate.gov XML (legislative/LIS/roll_call_lists/) for Senate votes — joined into one collection. While senate.gov refuses our requests, newly cast Senate votes may come from Voteview, labelled source 'voteview' with a source_citation and a source_note (the Senate's question text and document links are missing), and are replaced by the senate.gov record when the Senate answers again. Captures roll-call (recorded) votes only — voice votes and unanimous-consent passages aren't roll calls and don't appear here. v1A returns vote-level metadata: chamber, roll call number, vote type, result, the legislation being voted on (linked via bill_id), and links to the Clerk's authoritative XML data. Per-member positions (yea/nay/present/not voting per bioguide_id) live in the XML at source_data_url; agents fetch that directly when they need member detail. v1.1 will add a separate roll_call_member_votes tool/ collection for queryable per-member positions. Vote identifiers are stable composite keys: '{chamber}-{congress}- {session}-{rcNumber}', e.g., 'house-119-1-240' or 'senate-119-1-15'. Common vote_type values: 'Yea-And-Nay' (regular recorded vote), '2/3 Yea-And-Nay' (suspension of rules, requires 2/3 majority), 'Recorded Vote', 'Quorum'. Common result values: 'Passed', 'Failed', 'Agreed to', 'Rejected', 'Motion Agreed To', 'Motion Failed'. When a vote is on a bill, legislation_type + legislation_number are populated and bill_id is set to the composite key — use that to join to get_bills. For procedural votes (motion to recommit, motion to adjourn, etc.), those fields may be empty. Amendment + Senate detail: House votes ON AN AMENDMENT carry amendment_number, amendment_author (sponsor + label), and amendment_type (e.g. 'HAMDT'). Senate votes carry vote_title (descriptive title — e.g. a confirmation or motion), measure (the specific measure a question references, e.g. 'S.Amdt. 5740'), and en_bloc_matters[] (one {issue, question, result} per matter when a batch of nominations is decided en bloc). All are empty / [] where not applicable.

Returns entries from the US Consolidated Screening List (CSL) — the combined feed of twelve federal export-screening lists. Use this when the user asks about: whether a company or person is on a US screening / sanctions / denied-party list (KeyVex republishes the lists; it is not a screening service), BIS Entity List members, Military End User designations, or to add a 'restricted party' flag to a federal contractor or foreign agent. The CSL unifies twelve lists (filter via source_short): SDN — Specially Designated Nationals (Treasury/OFAC) EL — Entity List (Commerce/BIS) DPL — Denied Persons List (Commerce/BIS) MEU — Military End User List (Commerce/BIS) UVL — Unverified List (Commerce/BIS) CMIC — Non-SDN Chinese Military-Industrial Complex Companies (Treasury) CAP — Capta List (Treasury) DTC — ITAR Debarred (State) ISN — Nonproliferation Sanctions (State) MBS — Non-SDN Menu-Based Sanctions List (Treasury) PLC — Palestinian Legislative Council List (Treasury) SSI — Sectoral Sanctions Identifications List (Treasury) Broader than get_ofac_sdn — the SDN list is just one source here. For the OFAC-SDN deep view use get_ofac_sdn; to search every US list at once use this tool. Cross-source pairing: pair with get_federal_contracts to flag a contractor that also appears on a screening list, and with get_foreign_agents for the foreign-entity overlay. Each record carries name + alt_names, the source list, sanctions programs, addresses, distinct countries, and identification documents.

Returns SEC comment-letter correspondence: form UPLOAD (the SEC's letter TO the company — the questions) and CORRESP (the company's response). The Division of Corporation Finance sends these during filing reviews; they're released ~20+ business days after the review closes. Coverage 2005→present. Use this when the user asks about: whether a company is (or was) under SEC review, accounting-quality red flags before they become enforcement, the back-and-forth around an IPO registration, or to pair with fundamentals / insider activity ('were insiders selling while the SEC was asking questions?'). Reading a thread: filter by ticker or cik, sort date_filed asc — a review is an alternating UPLOAD/CORRESP chain; the final short UPLOAD is typically the 'review complete' letter. v1A is metadata-only: follow filing_index_url for the letter text. released_date is set on records captured from EDGAR's daily indexes (the dissemination day); older backfilled records carry only date_filed (the letter's own date) — dissemination day isn't recoverable historically and KeyVex never fabricates it. A comment letter is ROUTINE, not an accusation — most large filers get reviewed on a cycle (Sarbanes-Oxley §408 requires review at least every 3 years). Signal comes from thread LENGTH, topic, and recency, which agents judge from the letter text. Pure-publisher posture: EDGAR index records as published.

Returns SEC Fails-to-Deliver (FTD) rows — daily settlement failures by ticker / CUSIP / date. Each row is one ticker on one settlement date where a clearing-member's short sale FAILED to deliver shares. Signal value: persistent FTDs are a contrarian short-squeeze leading indicator. When the daily FTD quantity spikes on a ticker, it often means naked short pressure overwhelming locate supply or settlement / locate mechanism breaking down. The Reg SHO Threshold Securities list (FTDs > 0.5% of issued shares for 5+ consecutive days) is a derived view; this tool exposes the underlying daily data. Source: SEC bi-monthly cnsfails<YYYYMM><a|b>.zip files at sec.gov/files/data/fails-deliver-data/. Published ~1 week after each half-month settlement period. Coverage: every U.S.-listed security with a recorded settlement failure during the period. Killer query patterns: - Daily FTD history for a ticker: ticker='GME' + sort_by='settlement_date' - Largest FTDs this month: min_value=1000000 + sort_by='fail_value' - Squeeze setup candidates: min_quantity=100000 + recent dates - Look-up by CUSIP: cusip='B6S7WD106' (foreign issuers, complex names) Derived field: fail_value = quantity_fails × price (dollar magnitude of the failure on that day). Reference price comes from the SEC's posted value at settlement. Note: FTDs are bi-monthly batch-published, not real-time. SEC releases each half-month batch (cnsfails<YYYYMM>a = days 1-15, b = 16-end) roughly 2-4 weeks AFTER that half-month period closes, so the most recent settlement date can be 2-4 weeks behind today (e.g. in mid-June the latest published batch is first-half-May, with settlement dates through ~May 15). That apparent lag is the SEC publish cadence, not a KeyVex freshness gap.

Returns SEC Schedule TO filings — public tender offer disclosures. Use this when the user asks about: who's bidding to acquire company X, what M&A offers are in flight, share buyback announcements, amendments to existing tender offers (price increases / extensions), or to pair with 13D activist stakes for the 'stake → bid' story. Source: SEC EDGAR full-text search. Forms covered: SC TO-T (third- party tender offer — someone outside the company bidding for shares), SC TO-T/A (amendments), SC TO-I (issuer tender offer — company buying back its own shares), SC TO-I/A (issuer amendments). v1 returns filing metadata only — bidder + target + form type + filing date + URL. Offer price, shares sought, and expiration date live inside the HTML attachment at primary_document_url; agents follow that URL to read the substantive terms. Amendment filings share the same target/bidder/file_number as the original offer; use file_number to group an amendment chain. Pure-publisher posture: KeyVex does not derive 'likely to close' or 'expected premium' signals. The data here is what was filed, no more.

Returns Treasury security auctions — Bills (≤1yr), Notes (2-10yr), Bonds (20-30yr), TIPS (inflation-protected), and FRNs (floating-rate). Each record is one CUSIP issuance with announcement metadata + post- auction results. Key signal fields agents care about: - bid_to_cover_ratio: demand. >2.5 strong, <2.0 weak. - high_yield / average_yield: market clearing rate. - direct_bidder / indirect_bidder breakdowns: domestic vs foreign demand. - soma_holdings + soma_included: Fed System Open Market Account allocation. A live measure of Fed QE/QT activity on each issue. Records have a two-stage lifecycle: announcement (results fields null) → post-auction (full results populated). Idempotent saves on cusip + auction_date overwrite cleanly when results publish. Security types: 'Bill', 'Note', 'Bond', 'TIPS', 'FRN', 'CMB' (cash- management bill). Use security_type filter to focus on one term group. Note: Treasury reports TIPS and FRNs under security_type Note/Bond with an inflation-indexed / floating-rate flag (not as their own type); filtering security_type:'TIPS' or 'FRN' here resolves to those flags for convenience.

Identifier-driven cross-collection fan-out search. Pass one or more entity identifiers — ticker, bioguide_id, company_cik, recipient_uei, company_name, or cusip — and this tool queries every collection where that field is indexed, returning results grouped by source in one envelope. Congressional-trade and annual financial disclosure results come from STOCK Act and Form 278 filings. The same filings are published free as news at https://keyvex.com/disclosures under 5 U.S.C. § 13107(c). Use this for high-level 'tell me everything about X' questions before drilling into specific source tools. Replaces 6-10 sequential tool calls with a single fan-out. Identifier coverage: - ticker → 13 collections (company_profile, insider_trades, institutional_holdings, congressional_trades, planned_insider_sales, initial_ownership_baselines, activist_ownership, material_events, proxy_filings, xbrl_fundamentals, tender_offers, registration_statements, nport_holdings) - bioguide_id → 2 collections (congressional_trades, annual_financial_disclosures). In both, a record's party is the party the member held on the date of the record (a trade's transaction date, a disclosure's filing date); for a date outside the member's terms in office, the party of their nearest term (the last one before that date, or the first one after it). - company_cik → 11 collections (company_profile, insider_trades, planned_insider_sales, initial_ownership_baselines, activist_ownership, material_events, proxy_filings, xbrl_fundamentals, private_placements, registration_statements, nport_filings) - recipient_uei → 1 collection (federal_contracts) - company_name → 5 name-keyed collections (federal_contracts, enforcement_actions, consumer_complaints, product_recalls, fda_approvals) AND auto- resolves to ticker + company_cik via EDGAR's catalog, cascading into every ticker/CIK adapter above. The unlock for 'tell me everything about Wells Fargo' hitting up to 20 collections in one call. (lobbying_filings is excluded from the fan-out — its 51K-record substring scan is too slow for parallel federation; call get_lobbying_filings directly instead.) enforcement_actions here returns ENFORCEMENT only, as get_enforcement_actions does by default; call get_enforcement_actions with category="all" for the agencies' general news too. When a regulator's site refuses KeyVex (CFTC since 2026-10-07), the freshness block says which records are missing. - cusip → 4 collections (institutional_holdings, activist_ownership, nport_holdings, treasury_auctions) Multiple identifiers can be combined to narrow each source's query (e.g., ticker + bioguide_id will filter congressional_trades by both). Name resolution: when only company_name is supplied, the resolved ticker and CIK come from EDGAR's company_tickers_exchange.json (US-listed names only). Foreign-only or private companies won't resolve to a ticker, but name-keyed collections (CFPB, lobbying, etc.) still receive the substring filter directly. Performance note: company_name fan-out includes substring-filtered collections (lobbying_filings 51K+, federal_contracts) which scan a 5K- record window per source. The full cascade can take 10-40s depending on Firestore region. When latency matters more than coverage, pass `sources` to whitelist only the ticker/CIK-keyed adapters (typically <1s total). Per-source result count is capped via per_source_limit (default 5, max 50). One slow or failing source returns an error block instead of blocking the rest — check sources_queried vs sources_with_results to see what landed. When you need full result counts and richer filters on one specific collection, call that collection's dedicated tool directly afterward.