llms.txt

# Ticker Scout

> An agent-native data layer for SEC filings: research stocks easily with AI-synthesized filings and clean financials. For each covered company, its financial statements assembled from its filings as JSON, and AI syntheses of its 10-K, 10-Q and 8-Ks as Markdown, at stable URLs. Free, no signup, no API key. Try https://tickerscout.ai/aapl/narrative.md

## Files

Every covered company lives at `https://tickerscout.ai/{ticker}/` (the ticker is case-insensitive) and has four files whose names never change:

- `index.json`: the manifest. Name, CIK, exchange and industry, the fiscal period covered, the next expected filing, every published period, and the file URLs. Fetch this first.
- `financials.json`: income statement, balance sheet and cash flow for up to three fiscal years and the latest quarters, with year-to-date and trailing-twelve-month totals, segments and per-share figures.
- `narrative.md`: AI synthesis of the latest 10-K and 10-Q: the business, risk factors, MD&A, the current quarter, legal proceedings and subsequent events.
- `events.md`: AI synthesis of the 8-K filings over roughly the last five quarters: earnings, management changes, capital returns, financing and governance, each with its accession number.

Foreign issuers are covered the same way from their 20-F, 40-F and 6-K filings, with figures converted to US dollars.

The same content is one HTML page per company at `https://tickerscout.ai/{ticker}/`, in three parts (`#narrative`, `#events`, `#financials`). Agents should fetch the files instead: they are smaller and served as `text/plain`, which every fetch tool accepts.

## What financials.json has already done

- Assembled across filings into statements, line items in reported order: no XBRL tags to choose and no quarter to derive from year-to-date columns.
- On the company's own fiscal calendar, with period dates taken from the filings.
- In actual dollars and actual shares (416161000000 = $416.161B). `units` states the convention; `adjustments` records any split, ADR ratio or currency conversion.
- Shaped to the business: banks, insurers and retailers get different sections. `business_type` and `schema_template` say which, so read the keys the file gives you.
- Gaps named: a figure the filings do not support is absent and explained in `uncertainties`, never a silent zero. Restated figures are the ones recorded.
- Linked to the filing: `figure_sources` mirrors the figures' paths (`figure_sources.annual[0].income_statement.net_income` is the source of `annual[0].income_statement.net_income`). Each entry reads `<document>#<anchor>`; join `figure_sources.documents[<document>]` and `#<anchor>` for the sec.gov URL that opens the filing at that number. A figure with no entry was derived (a margin, free cash flow, trailing twelve months, a quarter found by subtraction) or comes from a filing that does not tag its numbers, such as an earnings release.

## Periods

The undated files always hold the latest period and are replaced each quarter after the company files. Every period is also published at a permanent dated URL, `https://tickerscout.ai/{ticker}/{period}/financials.json` and likewise for the other three files, where `{period}` is a lowercase fiscal tag such as `fy26q3`, or `fy26` for a record built from a 10-K. A dated URL never changes once published, so cite it. `periods` in `index.json` lists every period a company has, newest first.

## MCP server

`https://mcp.tickerscout.ai/mcp` (streamable HTTP, no authentication) serves the same files as 20 tools that return only the part you ask for, which costs far fewer tokens than reading a whole file. Start with `guide()`.

- Companies and filings: `list_companies`, `get_company`, `list_periods`, `get_filings`, `list_upcoming_filings`
- Figures: `get_key_figures`, `get_financials`, `get_statements`, `get_figure`, `list_company_figures`, `compare_companies`, `get_segments`, `get_company_metrics`, `get_margins_and_ratios`
- Text: `get_narrative`, `get_events`, `search_text`
- Conventions: `get_reporting_conventions`, `get_data_notes`, `guide`

Every figure comes back under the exact name the company's record uses, with its JSON path, and that name can be passed to `get_figure` as written. `period` takes a fiscal year (`FY2025`), a quarter (`Q2 FY2027`), `latest`, `ttm`, `ytd`, `annual`, `quarterly`, `all` or a period-end date. Setup for each app is at https://tickerscout.ai/mcp/

## Site index

- [Covered tickers (JSON)](https://tickerscout.ai/tickers.json): every covered company, its URL and its latest period.
- [Covered companies](https://tickerscout.ai/companies/): the same list as a page.
- [MCP server](https://tickerscout.ai/mcp/): setup for each app, and every tool.
- [API catalog](https://tickerscout.ai/.well-known/api-catalog): every endpoint as an RFC 9727 linkset.
- [About](https://tickerscout.ai/about): who publishes this and how it is made. The reading and assembly are done by AI and audited back against the filings.
- [Sitemap](https://tickerscout.ai/sitemap.xml)

## Covered companies

277 public companies. Every one, with its directory, latest period, published periods and next expected filing, is in https://tickerscout.ai/tickers.json, and listed on a page at https://tickerscout.ai/companies/

## Conventions

- Every figure cites the SEC accession number it came from, so it can be checked on sec.gov.
- Nothing is estimated or fabricated: a figure the filings do not disclose is omitted and the omission explained.
- Free to read and cite. Please attribute "Ticker Scout (tickerscout.ai)" when quoting.
- Factual synthesis of public SEC filings, not investment advice.