| name | edgarparse |
| description | Use this skill whenever the user asks about historical financial data, earnings, revenue, profit, debt, cash flow, or any financial metrics for a publicly traded US company — even if they don't mention "edgarparse" or "API." Trigger proactively on any question that could be answered with SEC filing data. |
EdgarParse API Skill
Fetch structured historical financial data for US public companies via the EdgarParse API, which parses SEC EDGAR 10-Q and 10-K filings into clean JSON — no scraping, no multiplier guessing.
API key: https://edgarparse.com/login — free tier covers 100 major tickers, up to 120 quarterly or 10 annual periods.
Endpoints
Base URL: https://api.edgarparse.com/v1
| What you need | Endpoint |
|---|
| Revenue, profit, EPS | GET /v1/tickers/{ticker}/income |
| Assets, debt, equity, cash | GET /v1/tickers/{ticker}/balance |
| Operating/investing/financing cash | GET /v1/tickers/{ticker}/cashflow |
| Check free-tier tickers | GET /free-tickers |
Add ?period=quarterly (default) or ?period=annual, and &api_key=YOUR_KEY.
Response
{
"ticker": "AAPL",
"period": "quarterly",
"periods": ["2024-09-30", "2024-06-30", "2024-03-31"],
"line_items": [
{
"concept": "us-gaap:Revenues",
"label": "Revenue",
"values": [94930000000, 85777000000, 90753000000],
"format": "currency",
"is_bold": false
}
]
}
periods[i] pairs with values[i] — zip them to get a time series. format is "currency", "per_share", or "shares". Values are raw dollars (not in millions).
Steps
- Map to ticker — if the user named a company, resolve it (Apple → AAPL, Tesla → TSLA).
- Pick one endpoint — income for P&L questions, balance for assets/debt, cashflow for cash generation. Only call multiple if the user needs data that spans statements.
- Fetch quarterly by default — switch to annual only if the user asks or if quarterly data would be too granular.
- Present as a table — zip
periods + values for the relevant line items, format large numbers (e.g. $94.9B), and highlight the trend.
Avoid fetching all three endpoints speculatively — it's slower and most questions need only one.