SEC EDGAR Filings & Fundamentals API — RAG-Ready

Turn the U.S. SEC EDGAR database into embeddings-ready, citation-rich data for AI / RAG pipelines and finance agents. Give it tickers (or CIKs, or a full-text query); get back clean Markdown chunks of every 10-K, 10-Q, and 8-K — split by SEC Item section, token-budgeted, deep-linked — plus XBRL fundamentals (Revenues, Net Income, EPS, Assets, cash flow) attached to each filing.

No API keys. Keyless EDGAR endpoints only, with a polite 10 req/s token-bucket limiter and a descriptive User-Agent, exactly as the SEC requires.

What it does

1. Ticker → CIK resolution via the canonical company_tickers.json.
2. Filing listing & filtering by form type and date window from the EDGAR submissions API.
3. Full-text search across all of EDGAR via efts.sec.gov (optional fullTextQuery).
4. Primary document fetch + clean HTML → Markdown (Readability + Turndown/GFM — financial tables survive intact).
5. SEC Item segmentation (Item 1A. Risk Factors, Item 7. MD&A, 8-K Item 2.02, …) so every chunk carries an Item anchor.
6. Structure-aware, token-budgeted chunking with overlap and headingsPath.
7. XBRL fundamentals per company via the companyconcept / companyfacts API.

Why use it (SEO)

• SEC EDGAR API for AI without writing a scraper.
• 10-K / 10-Q / 8-K filings as RAG chunks with working citations and deep links.
• XBRL fundamentals — Revenues, Net Income, EPS, Total Assets, operating/investing/financing cash flow.
• Drop-in formats for LangChain, pgvector / vector DB bulk upsert, or generic JSONL.
• Built for financial analysis agents, equity research copilots, and compliance/risk tooling.

Input

Field	Type	Default	Notes
tickers	string[]	—	e.g. ["AAPL","MSFT"]
ciks	string[]	—	raw CIKs, zero-padded internally
filingTypes	string[]	["10-K","10-Q","8-K"]	EDGAR form types
since / until	string	—	ISO date window on filingDate
fullTextQuery	string	—	phrase search across all EDGAR
maxFilingsPerCompany	int	25	per-company cap
maxFilingsTotal	int	0	global cap (0 = unlimited)
extractChunks	bool	true	fetch + chunk primary docs
chunkSize / chunkOverlap	int	512 / 64	token budget
splitByItem	bool	true	segment on SEC Item boundaries
includeFundamentals	bool	true	attach XBRL facts
xbrlConcepts	string[]	built-in ~20	override the us-gaap allowlist
xbrlSource	enum	companyconcept	companyconcept or companyfacts
outputFormat	enum	chunks-jsonl	chunks-jsonl, langchain, jsonl-bulk, filings-only
userAgentEmail	string	—	contact email for the SEC User-Agent

At least one of tickers, ciks, or fullTextQuery is required.

Output

Two record kinds land in the dataset:

• recordType: "chunk" — the primary RAG artifact. Markdown text + headingsPath + a citation block (ticker, CIK, form, accession, filing date, Item code/title, deep link, and a human-readable citation string).
• recordType: "filing" — one per filing, carrying fundamentals (latest observation per GAAP concept) and provenance. outputFormat: "filings-only" emits only these.

Default GAAP concepts (us-gaap)

Revenues, RevenueFromContractWithCustomerExcludingAssessedTax, CostOfRevenue, GrossProfit, OperatingIncomeLoss, NetIncomeLoss, EarningsPerShareBasic, EarningsPerShareDiluted, ResearchAndDevelopmentExpense, Assets, AssetsCurrent, Liabilities, LiabilitiesCurrent, StockholdersEquity, CashAndCashEquivalentsAtCarryingValue, LongTermDebtNoncurrent, NetCashProvidedByUsedInOperatingActivities, NetCashProvidedByUsedInInvestingActivities, NetCashProvidedByUsedInFinancingActivities, PaymentsToAcquirePropertyPlantAndEquipment.

Endpoints used (all keyless)

• https://www.sec.gov/files/company_tickers.json
• https://data.sec.gov/submissions/CIK##########.json
• https://efts.sec.gov/LATEST/search-index
• https://www.sec.gov/Archives/edgar/data/.../index.json + primary documents
• https://data.sec.gov/api/xbrl/companyconcept/... and .../companyfacts/...

Local development

npm install
npm run build
apify run   # uses .actor/INPUT.json