SEC EDGAR Full-Text Filing Search

The SEC publishes every public-company filing since 1994 through its SEC EDGAR full-text search index (efts.search). This SEC EDGAR actor queries that index by free-text phrase, CIK, form type, and date range and returns each filing hit as a structured record — accession number, CIK, ticker, SIC code, filing date, document URL.

Built for the workflows where you need to answer questions like:

• "Show me all 10-K filings mentioning insider trading policy filed in Q1 2024."
• "Pull every Form 4 insider-trading report from Apple (CIK 320193) in 2025."
• "Find recent 8-K disclosures for any issuer in SIC code 737 (software / data processing)."
• "Get every recent exhibit from any issuer that includes the words going concern in the body."

No API keys, no OAuth, no captchas. Polite-pool only — a descriptive User-Agent and a 120ms inter-request sleep are baked in.

What it returns

One record per SEC EDGAR filing hit. Real sample from a 2026-09-01 smoke run on input {"ciks":"0000320193","forms":"10-K","dateRange":"custom","startdt":"2023-01-01","enddt":"2024-12-31","maxItems":2} (run id 4e4PTHh1dzMeet2cG, dataset zPjRUG11ywzefgVG8):

{
  "id": "0000320193-24-000123:aapl-20240928.htm",
  "accessionNumber": "0000320193-24-000123",
  "accessionUrl": "https://www.sec.gov/cgi-bin/browse-edgar?action=getcompany&CIK=0000320193&type=10-K&dateb=&owner=include&count=40",
  "form": "10-K",
  "fileType": "10-K",
  "fileDescription": "10-K",
  "fileDate": "2024-11-01",
  "periodEnding": "2024-09-28",
  "cik": "0000320193",
  "allCiks": "[\"0000320193\"]",
  "displayNames": "[\"Apple Inc.  (AAPL)  (CIK 0000320193)\"]",
  "tickers": "[\"AAPL\"]",
  "sic": "3571",
  "sicDescription": "Electronic Computers",
  "bizState": "CA",
  "bizLocation": "Cupertino, CA",
  "fileNum": "001-36743",
  "filmNum": "241416806",
  "documentUrl": "https://www.sec.gov/Archives/edgar/data/320193/000032019324000123/aapl-20240928.htm",
  "score": null,
  "sourceQuery": "(none) + ciks=0000320193 + forms=10-K + dateRange=custom + startdt=2023-01-01 + enddt=2024-12-31",
  "fetchedAt": "2026-09-01T23:47:13.000Z"
}

Input

Matches the actor's INPUT_SCHEMA.json exactly:

Field	Required	Description
query	conditional	Free-text search. Supports phrase queries ("insider trading") and AND/OR operators. Required if ciks is empty.
ciks	conditional	Comma-separated CIK numbers (10-digit zero-padded or trimmed, e.g. 0000320193 or 320193). Required if query is empty.
forms	no	Comma-separated form-type filter (e.g. 10-K,10-Q,8-K,4). Empty for all forms.
dateRange	no	EDGAR built-in preset. Allowed: custom, today, yesterday, last-week, last-month, last-quarter, last-year, 1d, 5d, 30d, 90d, 1y. Use custom with startdt/enddt.
startdt	conditional	YYYY-MM-DD lower bound. Only when dateRange='custom'.
enddt	conditional	YYYY-MM-DD upper bound. Only when dateRange='custom'.
maxItems	no	Max records to return (1–100, default 10). EDGAR /search-index caps page size at 100.

Validation: at least one of query or ciks must be non-empty. The actor fails fast at startup if both are missing.

Example inputs:

# Query mode: "insider trading" in 10-K filings, Q1 2024
echo '{"query":"\"insider trading\"","forms":"10-K","dateRange":"custom","startdt":"2024-01-01","enddt":"2024-03-31","maxItems":5}' > .actor/input.json

# CIK mode: every Form 4 from Apple in 2025
echo '{"ciks":"0000320193","forms":"4","dateRange":"custom","startdt":"2025-01-01","enddt":"2025-12-31","maxItems":10}' > .actor/input.json

# Combined: software industry (SIC 7372) + "data breach"
echo '{"query":"data breach","forms":"8-K","dateRange":"last-month","maxItems":10}' > .actor/input.json

How it works

The actor hits https://efts.sec.gov/LATEST/search-index?q=<query>&ciks=<ciks>&forms=<forms>&dateRange=<preset> (or with startdt/enddt for dateRange=custom), parses the JSON response, and projects each hit into a flat record. List-valued fields (allCiks, displayNames, tickers) are JSON-encoded as strings because Apify's output schema v1 only renders string-typed fields. Pagination beyond maxItems=100 requires an external loop — intentional, because EDGAR caps page size at 100.

Run

make install
make run        # local smoke against INPUT_SCHEMA.json
make push       # build + push to Apify
make run-prod   # cloud smoke on the published build

Limits and gotchas

• Polite-pool rate limit. SEC guidance: "no more than 10 requests per second" with a descriptive User-Agent. The actor sends one request per run by default.
• No proxy. This is a keyless public endpoint. SEC blocks anonymous datacenter IPs at peak hours; consumer traffic from Apify's IP pool works fine for typical workloads but heavy commercial usage should consider a polite-pool agreement with SEC.
• SIC subset is partial. The actor embeds a ~250-entry subset of SEC's SIC table covering finance, real estate, manufacturing, retail, services, and government. Less-common SIC codes will return sicDescription=null — consumers can resolve via SEC's published SIC list.
• Free-text quirks. EDGAR uses Elasticsearch match-phrase semantics. Bareword queries match across all token positions; quoted strings anchor to phrase order; capitalisation is normalised. Wildcard * returns an empty result set — use forms + date filters to scope instead.
• Score field. EDGAR's score is null for CIK-only queries (the index does not compute relevance when the query is structurally constrained). For text queries it returns a relevance number.

License

SEC EDGAR is a US SEC public service; filing metadata is in the public domain.