<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:dc="http://purl.org/dc/elements/1.1/">
  <channel>
    <title>DEV Community: Jaroslav Šmarda</title>
    <description>The latest articles on DEV Community by Jaroslav Šmarda (@jaroslav_marda_9f25d4d9c).</description>
    <link>https://dev.to/jaroslav_marda_9f25d4d9c</link>
    <image>
      <url>https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F4083355%2F70d11f08-ebaf-444f-bacf-228cb5c92b20.jpg</url>
      <title>DEV Community: Jaroslav Šmarda</title>
      <link>https://dev.to/jaroslav_marda_9f25d4d9c</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/jaroslav_marda_9f25d4d9c"/>
    <language>en</language>
    <item>
      <title>Free, Unrestricted, and Almost as Easy: Reading SEC Filings with a Prompt</title>
      <dc:creator>Jaroslav Šmarda</dc:creator>
      <pubDate>Wed, 19 Aug 2026 11:06:51 +0000</pubDate>
      <link>https://dev.to/jaroslav_marda_9f25d4d9c/free-unrestricted-and-almost-as-easy-reading-sec-filings-with-a-prompt-44ai</link>
      <guid>https://dev.to/jaroslav_marda_9f25d4d9c/free-unrestricted-and-almost-as-easy-reading-sec-filings-with-a-prompt-44ai</guid>
      <description>&lt;p&gt;Python for programmers, prompts for data analysts.&lt;/p&gt;

&lt;p&gt;Every article in this series has leaned on one server: EODHD. But the underlying numbers — revenue, profit, shares outstanding — don't originate there. They come from a company's own filings to the SEC, published for free, for anyone, the moment they're submitted.&lt;/p&gt;

&lt;p&gt;So here's a natural question: could we skip the middleman entirely, and just ask the source?&lt;/p&gt;

&lt;p&gt;Turns out, mostly yes.&lt;/p&gt;

&lt;p&gt;New here? The Setup covers everything you need before your first prompt.&lt;/p&gt;

&lt;p&gt;Ask the source directly&lt;br&gt;
🟧 PROMPT&lt;/p&gt;

&lt;p&gt;Fetch data.sec.gov/api/xbrl/companyfacts/CIK0001018724.json for Amazon and tell me their most recent annual revenue.&lt;/p&gt;

&lt;p&gt;Amazon's CIK (Central Index Key) is 1018724 — padded to 10 digits for the URL. Fetching that page returns a single JSON document containing every financial fact Amazon has disclosed to the SEC since it started filing electronically: shares outstanding, accounts payable, revenue, and hundreds more, each as a full time series stretching back years.&lt;/p&gt;

&lt;p&gt;No login, no API key, no special header required to read it — the page opened on the first request.&lt;/p&gt;

&lt;p&gt;No MCP server. No Docker. Not even a signup. The exact same government database that powers every "10-K" citation you've ever seen in a finance article is sitting behind a plain URL, and Claude can just fetch it — the same way it's fetched news articles and pricing pages throughout this series.&lt;/p&gt;

&lt;p&gt;Pull a real number&lt;br&gt;
🟧 PROMPT&lt;/p&gt;

&lt;p&gt;From that data, what was Amazon's most recent quarterly accounts receivable balance?&lt;/p&gt;

&lt;p&gt;The most recent figure in the file: $88.09 billion, for the quarter ending June 30, 2026, filed August 1, 2026. The same concept — "AccountsReceivableNetCurrent" — stretches back to 2008, so you can trace the exact same line item across nearly two decades in one file.&lt;/p&gt;

&lt;p&gt;That's the pattern for anything company-wide: revenue, cash, debt, shares outstanding — one concept name, one flat time series, every quarter and every year a company has filed, going back as far as their electronic records exist. No parsing a PDF, no scraping a table out of a press release. It's already structured, and it's already sitting there.&lt;/p&gt;

&lt;p&gt;Ask for the one thing it can't give you&lt;br&gt;
🟧 PROMPT&lt;/p&gt;

&lt;p&gt;From that same file, break out Amazon's revenue by segment — AWS, North America, and International.&lt;/p&gt;

&lt;p&gt;Looking through the file, revenue is there — but only as one number per period, the company-wide total. There's no "AWS revenue" or "North America revenue" concept in this list. The segment breakdown genuinely exists in Amazon's 10-K — it's just not in this particular document.&lt;/p&gt;

&lt;p&gt;That's the wall. Everything company-wide — one number per concept, per quarter, forever — comes through cleanly. But a segment split isn't a company-wide fact; it's a number tied to which part of the business it belongs to, and that extra dimension doesn't travel through this endpoint. The data hasn't vanished. It's just filed under a different door than the one we just walked through.&lt;/p&gt;

&lt;p&gt;A word of honesty&lt;br&gt;
Two things worth being precise about, and they cut in opposite directions.&lt;/p&gt;

&lt;p&gt;First, the good news is real: anything company-wide — revenue, cash, debt, shares outstanding, hundreds of concepts — comes through a plain URL, no signup, no key, no rate-limit dance. That's rarer than it sounds; most financial APIs gate at least some of this behind a paid tier. SEC EDGAR doesn't, because it's not a product — it's a public record.&lt;/p&gt;

&lt;p&gt;Second, "no MCP server needed" only holds for exactly this kind of question — one concept, one company, over time. The moment you want something structurally different, like a segment split, you've left the territory this endpoint covers. That's not a flaw in the API; it's a reminder that "free and unrestricted" and "covers everything you might want" are different promises, and this only makes the first one.&lt;/p&gt;

&lt;p&gt;So: a genuinely good tool for the numbers a company reports about itself as a whole — and a dead end, at least here, for the numbers it reports about its parts. As always: not investment advice.&lt;/p&gt;

&lt;p&gt;Final Thoughts&lt;br&gt;
This one sits outside the main numbered series — a detour to check something the series has leaned on all along without ever looking directly at it: the government record every EODHD number ultimately traces back to.&lt;/p&gt;

&lt;p&gt;Two prompts got us company-wide facts going back nearly two decades, no setup beyond a URL. A third hit a wall the moment we asked for something structural instead of company-wide.&lt;/p&gt;

&lt;p&gt;So here's where this leaves things: SEC EDGAR is as open as data gets — free, unrestricted, no key, no rate-limit dance — and it will answer almost any question about a company as a whole. Ask about its parts, and you've asked a different kind of question than this endpoint was built to answer. That's not a flaw to work around. It's the shape of the thing, worth knowing before you go looking for it.&lt;/p&gt;

</description>
      <category>stockmarket</category>
      <category>claude</category>
    </item>
  </channel>
</rss>
