Skip to content

sec-edgar

Companion skill for Stefano Amorelli's sec-edgar-mcp (not vendored because it is AGPL). Covers CIK resolution, 10-K/10-Q/8-K retrieval, XBRL financial-statement facts, and Form 3/4/5 insider trades. It sets an iron law of no claim without sample inspection, requires a real SEC User-Agent, and warns that filing date and event date can differ for 8-Ks and that XBRL history is restated, so accession numbers must be pinned for replication. Sends Compustat/CRSP panel work to the wrds skill.

Category: data-handling
Field: accounting
License: MIT (repo LICENSE, "Copyright (c) 2026 John Barrios"). The vendored third-party skills keep their own terms: the Anthropic document skills say "Proprietary. LICENSE.txt has complete terms", and the K-Dense skills carry per-library licence lines.
Updated: 2026-07-23
Stages: data-acquisition

Barrios Skills — John Barrios's curated workflow for economists and accountants. Prioritize reproducible empirical work, clear identification language, and journal-ready output.

Use Stefano Amorelli's sec-edgar-mcp (built on edgartools) when the agent needs live filings.

Before querying EDGAR, the user must:

  1. Follow install/sec-edgar-mcp.md
  2. Set a real User-Agent: SEC_EDGAR_USER_AGENT="Name (email@institution.edu)" (SEC fair-access requirement)
  3. Prefer MCP tools over ad-hoc scraping when the server is registered

When to use WRDS instead: Compustat fundamentals panels, CRSP returns, CCM links, ExecuComp, ISS, TRACE/TAQ — use the wrds skill + wrds-mcp. EDGAR is for filing text, XBRL facts, 8-K events, and public insider Forms 3/4/5 when you do not need WRDS panel infrastructure.


SEC EDGAR for accounting & finance research

Iron law: no claim without sample inspection

Before asserting that a filing pull "worked":

  1. IDENTIFY the firm (CIK or ticker → CIK) and form type
  2. VALIDATE the accession / period you intended
  3. EXECUTE the pull
  4. INSPECT a sample (company name, filing date, section heading, one numeric fact)
  5. CITE the EDGAR URL in notes so numbers are auditable

Rationalization table — stop if you think

Excuse Reality Do instead
"Ticker is enough" Tickers recycle; CIK is stable Resolve CIK first, store it
"XBRL equals Compustat" Tag coverage and restatements differ Document source; reconcile if both used
"I'll scrape HTML tables" Layout breaks; XBRL/MCP is stabler Prefer XBRL facts or MCP financials tools
"Form 4 = Insider Trading database" Public EDGAR ≠ cleaned WRDS panels Use WRDS for research panels when available

Typical workflows

1. Company → filings

  • Resolve CIK from name or ticker
  • List recent 10-K / 10-Q / 8-K with filing dates
  • Extract a section (e.g. Item 1A, MD&A) only when needed — avoid dumping full HTML into context

2. Financial statement facts

  • Prefer XBRL-parsed balance sheet / income / cash flow from MCP
  • Record units, scale, and period end
  • Flag: never invent a ratio; compute from retrieved facts and show the formula

3. Insider trades (Form 3/4/5)

  • Pull transactions for a CIK and date window
  • Note: open-market vs award/grant; role of reporting person
  • For large-sample insider studies, prefer WRDS/Thomson when the user has access

Accounting / finance research notes

  • Identification language: filing date ≠ event date for some 8-Ks; use the relevant timestamp the paper needs
  • Restatements: XBRL history can change; pin accession numbers in replication folders
  • Fair access: respect SEC rate limits; never share another researcher's User-Agent email
  • Replication: save accession numbers + URLs alongside any extracted table

Output checklist

  • CIK recorded
  • Form type and accession (or filing URL) recorded
  • Sample row/section inspected
  • WRDS vs EDGAR choice justified if both could apply