SEC EDGAR Filings & Financial Corpus Scraper

Pull SEC EDGAR filings — 10-K, 10-Q, 8-K and 13F-HR — for any US-listed ticker and get each filing structured to JSON: filing metadata, sectioned document text, 10-K risk factors (Item 1A), 13F holdings tables, and an optional LLM-ready summary. Built on the SEC's official full-text search API — no API key, no third-party data vendor.

Why

EDGAR is the canonical source of US public-company disclosures, but the raw archive is hostile to analysis: filings are HTML documents with table-heavy layouts, and the full-text search UI is built for humans, not pipelines. This actor turns a ticker + form-type + date window into a clean corpus of JSON records — one per filing — ready for financial alt-data pipelines, sentiment/risk analysis, RAG corpora, or LLM fine-tuning datasets.

How it works

1. Ticker → CIK — resolves the ticker against the official SEC company list (company_tickers.json); falls back to display-name search for delisted or foreign-format tickers.
2. Full-text search — queries the official EDGAR full-text search JSON API (efts.sec.gov/LATEST/search-index) with the ticker, form types and date window, paginated at 100 hits per page, deduplicated by accession number, and filtered to each filing's primary document (exhibits are excluded; their URLs remain reachable via the filing index link).
3. Document extraction — fetches each primary document from the SEC Archives and converts HTML to clean text (scripts/styles stripped, tables preserved, entities decoded, whitespace normalised).
4. Structuring — splits documents into Item sections (Item 1., Item 1A., Item 7., Item 2.02, ...), extracts risk factors from 10-K/10-Q Item 1A, and parses 13F-HR XML InfoTables into holding records (issuer, CUSIP, value, shares, voting authority).
5. Summary — when summarize is enabled, emits a deterministic markdown summary per filing (metadata, section map, risk-factor excerpt, holdings preview) designed for direct LLM consumption — no external LLM call, no extra cost.

SEC fair-access policy is respected: a descriptive User-Agent (Darknezz Research admin@adrianhomelab.com) and polite pacing (~5 requests/second, well under the 10 rps limit) with 429 backoff.

Input

Field	Type	Default	Description
ticker	string	—	Stock ticker, e.g. AAPL, MSFT, TSLA (required)
formTypes	string[]	["10-K","10-Q","8-K","13-F"]	Form types to pull; 13-F maps to 13F-HR
period	string	empty	Date window: 2023-2026, 2024, or 2023-01-01:2026-12-31
summarize	boolean	true	Add the LLM-ready summary per filing
maxItems	integer	50	Max filings to extract (1–500)

Example:

{
  "ticker": "AAPL",
  "formTypes": ["10-K", "8-K"],
  "period": "2023-2026",
  "summarize": true
}

Output

One dataset item per filing:

• Metadata — accessionNumber, ticker, companyName, cik, formType, filingDate, periodEnding, event items (8-K), file details, documentUrl, sourceUrl (filing index).
• sections — array of {heading, item, text} for each Item section (per-section cap 60k chars; full text stays at the source URL).
• riskFactors — 10-K/10-Q Item 1A text (or any "Risk Factors" heading), empty for 8-K.
• holdings — 13F positions: nameOfIssuer, titleOfClass, cusip, value, shares, sharesType, putCall, investmentDiscretion, otherManager, votingSole/Shared/None. Mega-manager filings are capped at 20,000 positions with holdingsCount + holdingsTruncated flags.
• summary — LLM-ready markdown when summarize is true.
• charCount, scrapedAt.

Example record (Apple 8-K, fields trimmed):

{
  "accessionNumber": "0000320193-26-000045",
  "ticker": "AAPL",
  "companyName": "APPLE INC",
  "formType": "8-K",
  "filingDate": "2026-01-29",
  "items": ["2.02", "9.01"],
  "sections": [
    {"heading": "Item 2.02", "item": "2.02", "text": "Results of Operations and Financial Condition..."}
  ],
  "summary": "## Apple Inc. 8-K (2026-01-29)\n- Form: 8-K ...",
  "charCount": 4872,
  "scrapedAt": "2026-08-19T10:52:11.000Z"
}

Use cases

• Financial alt-data — monitor 8-K event items (2.02 results, 1.01 agreements, 9.01 exhibits) across a watchlist.
• Risk analysis — 10-K Item 1A risk-factor corpus for sector risk benchmarking or LLM risk-extraction pipelines.
• 13F tracking — institutional holdings snapshots (quarterly) for fund-flow and positioning analysis.
• Vertical AI corpus — clean, sectioned regulatory text for fine-tuning, RAG, or evaluation datasets.
• Fundamental screening — 10-Q/10-K MD&A text for narrative trend analysis over time.

Run from the API

curl -X POST "https://api.apify.com/v2/acts/darknezz~sec-edgar-filings-scraper/run-sync-get-dataset-items?token=YOUR_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"ticker":"AAPL","formTypes":["8-K"],"period":"2024-2026","summarize":true,"maxItems":5}'

Run from Python

import asyncio
from apify_client import ApifyClient

client = ApifyClient("YOUR-APIFY-TOKEN")

async def main():
    run = await client.actor("darknezz/sec-edgar-filings-scraper").call(
        run_input={"ticker": "MSFT", "formTypes": ["8-K"], "period": "2025-2026", "maxItems": 5}
    )
    for item in client.dataset(run["defaultDatasetId"]).iterate_items():
        print(item["accessionNumber"], item["formType"], item["filingDate"])
        print("items:", item.get("items"), "| sections:", len(item.get("sections", [])))

asyncio.run(main())

Pricing

Pay-per-event — $0.02 per filing extracted (result event) plus a $0.01 one-time run-start fee. You only pay for filings that are actually delivered to the dataset; an empty search costs nothing.

Worked examples: a 5-filing lookup costs $0.11 (5 × $0.02 + $0.01); a 100-filing sweep costs $2.01.

FAQ

Do I need an API key? No — EDGAR is a free public API. Only a descriptive User-Agent is required, which the actor sets automatically.

Rate limits? SEC's fair-access policy allows 10 requests/second; the actor stays at ~5 and backs off on 429s.

Why are exhibits missing from sections? Each filing yields its primary document only; exhibit files (EX-99.1 press releases, EX-10 contracts) are one click away via sourceUrl (the filing index).

Amendments? /A variants are included when the form-type filter matches (e.g. 10-K/A); use the exact form code to narrow.

What about delisted tickers? If the ticker is absent from the SEC company list, the actor falls back to display-name full-text search, which covers most historical filers.

Is the CIK lookup reliable? Yes — company_tickers.json is the SEC's own mapping; the zero-padded 10-digit CIK is used in search queries (the most common integration mistake).

Can I pull a 13F mega-manager like JPM? Yes, up to 20,000 positions per filing; larger filings are truncated with holdingsCount and holdingsTruncated flags so the record stays under Apify's item-size cap.

Limitations

• Primary document only — exhibits and multi-part filings beyond the primary document are not parsed (linked from sourceUrl).
• Per-section cap — sections over 60k characters are truncated; the full text remains at the source URL.
• 13F cap — positions beyond 20,000 per filing are truncated with explicit flags (records above Apify's 9 MB item cap would fail).
• Historical depth — the full-text index covers filings back to ~2001; pre-2001 documents are not searchable through this endpoint.