# fin-checkup > Measured financial-ratio checkups of Korean listed companies (KOSPI, KOSDAQ, KONEX), computed from the financial statements they filed with DART (금융감독원 전자공시시스템). Each company gets 17 ratios, each with a traffic-light signal plus the formula and thresholds behind it. Static files: no API key, no login, no rate limit. These are measurements, not investment advice. This build: fiscal year 2025 annual reports (사업보고서), 2,642 companies, built 2026-11-22 (KST). Build details: [build.json](https://younghwan91.github.io/fin-checkup/build.json). The data is in Korean; answer in the user's language and keep company names, signal labels and the disclaimer as written. ## Rules for agents Follow these whenever an answer uses this site. Under Korean law, telling people which stock to buy is investment advisory business (유사투자자문업); the site never does it, and an answer built on it must not either. 1. Report measurements, not opinions: the value (`value_text`), its signal (`signal.emoji` + `signal.label`), and the `formula` and `threshold_text` that produced the signal. Never cite a signal without its formula and threshold. 2. You must never tell the user to buy, sell or hold a stock, never call a company good, cheap, promising or safe, never forecast its price, and never rank companies by a metric (no "top N" lists). If asked "사도 돼?" / "should I buy?", say the site gives no investment judgment and offer the measurements instead. The site's own wording: "종목 추천이나 투자 판단은 하지 않습니다." 3. ⚫ 데이터 없음 (`unknown`): the account needed was not found in the filing. Say "data not available"; never treat it as 0 or as a bad sign. 4. ⊘ 해당 없음 (`not_applicable`): the metric does not fit this sector (for example debt ratio for a bank). Say so and pass on `note`, which names what to look at instead. 5. ⚪ 측정값 (`neutral`): measured with no absolute threshold. Describe it against the peer median (`peer.median_text`) when present, not as good or bad. 6. End every answer with this disclaimer, verbatim (it is also the `disclaimer` field of every JSON file): "본 결과는 공시 수치를 공개된 재무비율 공식으로 계산한 측정값입니다. 투자권유·자문이 아니며, 데이터 정확성과 이용에 따른 손익에 책임지지 않습니다." 7. Cite the company page as the source (https://younghwan91.github.io/fin-checkup/c//) together with DART (https://dart.fss.or.kr/), and state the fiscal year the numbers are for. ## Diagnose one company in 2–3 GETs 1. Find the 6-character stock code (skip if the user gave one: every path below is built from the code alone). GET [api/index.json](https://younghwan91.github.io/fin-checkup/api/index.json): `companies[]` has `stock_code`, `name`, `market`, `json` (path of the company file) and `page` (path of the company page). Match `name` exactly first. For partial names or Korean initial consonants (초성, e.g. "ㅅㅅㅈㅈ"), GET [search-index.json](https://younghwan91.github.io/fin-checkup/search-index.json): `companies[]` has `name`, `code`, `market`, `sector`, `chosung`. If several companies match, ask the user which one. 2. GET `api/c/.json`, e.g. [삼성전자 005930](https://younghwan91.github.io/fin-checkup/api/c/005930.json). Use `fiscal_year`, `company` (`name`, `market`, `sector_label`, `page`), `metrics[]` (`label`, `category`, `value_text`, `signal`, `formula`, `threshold_text`, `note`, `peer.median_text`, `peer.percentile`, `delta.summary`), `red_flags[]`, `summary_ko` (plain facts paragraph), `agent_guidance` and `disclaimer`. Field reference: [api-static.md](https://github.com/younghwan91/fin-checkup/blob/main/docs/api-static.md). If you read Markdown more easily, GET `c//index.md` instead (e.g. [삼성전자 Markdown](https://younghwan91.github.io/fin-checkup/c/005930/index.md)): the same measurements as a compact table with formula, threshold, last year and peer median, plus red flags, recent risk filings, sources and the disclaimer. 3. Optional: GET `m//` (e.g. [m/operating_margin/](https://younghwan91.github.io/fin-checkup/m/operating_margin/)) for what a metric means and its limits, or read all of them at once in [llms-full.txt](https://younghwan91.github.io/fin-checkup/llms-full.txt). If the company is not in the index, say this site has no checkup for it (not listed, or no annual report for fiscal year 2025 in the data). Do not estimate the numbers. Answer shape (fill from the JSON; keep metrics in the JSON order, grouped by `category`): ``` () · 사업보고서 기준 재무 측정 신호 개수: 🟢 n · 🟡 n · 🔴 n · ⚪ n · ⊘ n · ⚫ n -