EdgarTools MCP for SEC Filings

The question used to be "how do I pull data from SEC filings?" Now it's "why can't Claude do it for me?" Analysts, compliance teams, research desks — they've moved on. They're working inside AI tools, and they expect those tools to know what Apple reported last quarter. Instead they get hallucinated numbers, or a polite apology, or a suggestion to go check EDGAR themselves.

The answer is MCP — Model Context Protocol. An AI model only knows what's in its context window. It can't reach out and pull a 10-K but MCP is the standard that lets the scaffolding around the model do that on its behalf. It fetches real data from external sources and feed it back in. 

You're reading this because you're the developer, the quant, the data engineer who's going to wire that up. You're the person who connects the model to the data your team is already asking it about.

EdgarTools — the Python library for SEC EDGAR data — now ships an MCP server that does exactly that. 13 tools that give any MCP-compatible AI assistant — Claude Desktop, Claude Code, or your own internal application — structured access to every SEC filing, financial statement, and company disclosure. No API key, no paywall. The model stops guessing and starts pulling real data.

13 Tools, Organized by What You Want to Do

Most API wrappers expose endpoints. We designed tools around intent. You don't call get_filing_by_accession_number and then parse_xbrl_instance — you ask a question and the right tool handles the plumbing.

Discover

Tool	What it does
edgar_company	Start here. One call gets you profile, financials, filings, and ownership for any company.
edgar_search	Find companies by name or list filings by form type.
edgar_screen	Filter companies by industry, exchange, or state — using local data, zero API calls.
edgar_text_search	Full-text search across filing content via SEC EFTS.
edgar_monitor	See what was filed with the SEC in the last hour.

Examine

Tool	What it does
edgar_trends	Revenue, income, EPS time series with YoY and QoQ growth rates. XBRL-sourced.
edgar_compare	Side-by-side company comparison with automatic peer selection.
edgar_ownership	Insider transactions (Form 4) and institutional portfolios (13F).
edgar_fund	Mutual funds, ETFs, BDCs, money market funds — holdings, yields, performance.
edgar_proxy	Executive compensation and pay-vs-performance from DEF 14A proxy statements.

Analyze

Tool	What it does
edgar_trends	Revenue, income, EPS time series with YoY and QoQ growth rates. XBRL-sourced.
edgar_compare	Side-by-side company comparison with automatic peer selection.
edgar_ownership	Insider transactions (Form 4) and institutio\nal portfolios (13F).
edgar_fund	Mutual funds, ETFs, BDCs, money market funds — holdings, yields, performance.
edgar_proxy	Executive compensation and pay-vs-performance from DEF 14A proxy statements.

A cool tool is edgar_monitor. No other financial data MCP server offers a live filings feed. Claude can watch what's being filed right now and flag anything interesting — a new 8-K from a company you follow, a Form 4 insider purchase, a 13F portfolio disclosure.

Edgartools in Claude Desktop

To get started follow the MCP Setup Instructions. Once the server is running, you just talk to Claude. Here's what actually comes back — real data from real SEC filings.

Pre-Built SEC Filing Workflows

For common research patterns, the server ships with 7 multi-step prompts that chain tools together:

Prompt	What it does
due_diligence	Full company analysis — profile, financials, risks, insider activity
earnings_analysis	Earnings deep dive — latest 8-K, trends, peer comparison
industry_overview	Sector survey — screen companies, compare top players, identify trends
insider_monitor	Track insider buying and selling patterns
fund_analysis	Mutual fund or ETF deep dive — holdings, performance, fund family
filing_comparison	Compare filings across time or across companies
activist_tracking	Monitor SC 13D/G activist investor positions

These are templates, not black boxes. Claude follows them step by step, and you can interrupt, redirect, or drill deeper at any point.

What Makes This Different

There are other financial data MCP servers. Here's why this one is worth your attention:

Free. Not freemium, not "free tier with limits." Free. SEC EDGAR is a public resource funded by filing fees that public companies already pay. EdgarTools is MIT-licensed open source. The data flows from the SEC to your Claude session with no intermediary extracting rent.

No API key. No signup form. No OAuth flow. No usage dashboard. Set your EDGAR_IDENTITY and go.

Live data. edgar_monitor gives you a live feed of what's being filed right now. Most financial data services give you data from yesterday. The SEC publishes filings within minutes of receipt.

Structured, not raw. When Claude parses a 10-K through EdgarTools, it gets a typed TenK object with extracted financials, notes, and sections — not raw HTML that it has to interpret on the fly. Same for Form 4 insider transactions, 13F institutional holdings, DEF 14A proxy statements, and 30+ other form types.

Intent-based design. Tools match how analysts think, not how APIs are structured. "Tell me about Apple" is one tool call, not five.

Intelligent errors. If you ask for a company that doesn't exist, the server doesn't return a 404. It returns a structured error with suggestions: "Did you mean AAPL? Try edgar_search to find the right name."

The Bigger Picture

EdgarTools started as a Python library for parsing SEC filings. Then it became the most popular way to access SEC EDGAR data in Python — 2.3 million downloads and counting, used by hedge funds, academic researchers, and fintech teams. The MCP server is the next step: making that same data accessible to AI agents that are increasingly doing the analytical work that used to require a human with a Bloomberg terminal.

The SEC processes 500,000+ submissions a year. Every 10-K annual report, every 8-K material event, every Form 4 insider trade, every 13F institutional holdings disclosure — all of it is public data that tells you what's happening inside American companies. The hard part was never access. It was parsing XBRL, structuring financial statements, and making sense of it all. That's what EdgarTools does, and now Claude can do it too.

---

EdgarTools is an open-source Python library for SEC EDGAR filings, available under the MIT license. Install with pip install edgartools. The MCP server is registered in the official MCP registry. Full documentation, PyPI package, and source code are available. Contributions, bug reports, and GitHub stars are always welcome.