edgar-mcp

Name: EDGAR (SEC filings)
Author: mcpwright

SEC EDGAR filings, inside your agent. An MCP server that lets an LLM resolve companies, search filings, and pull recent securities offerings straight from the SEC — built on Anthropic's official mcp Python SDK.

All tools are read-only and hit public SEC endpoints (no API key required).

Status: 11 tools, working today (see below). Published on PyPI as mcpwright-edgar and in the official MCP Registry. See the roadmap for what's next.

Tools

Tool	What it does
`lookup_issuer(query, limit=10)`	Resolve a ticker or company name → CIK, legal name, tickers, exchange. Works for exchange-listed and private / non-exchange filers (Reg CF / Reg A issuers, funds).
`list_filings(cik_or_query, form_type=None, limit=20)`	An issuer's most recent filings, newest first. Optional form-type filter (e.g. `10-K`, `C`, `D`).
`search_filings(query, forms=None, date_from=None, date_to=None, limit=20)`	Full-text search across filing documents.
`get_recent_offerings(form="C", since=None, state=None, limit=20)`	Recent securities offerings, newest first — `form="C"` (Reg CF), `"D"` (Reg D), or `"A"` (Reg A — 1-A offering statements plus 253G2 offering circulars, where the per-share price and any "price walk-up" live), optionally filtered by issuer `state` (e.g. `"CA"`).
`get_filing(accession_or_url, cik=None)`	Open one filing: form, filing date, primary-document link, and every document in the filing.
`get_form_d_details(accession_or_url, cik=None)`	Parse a Form D (Reg D) raise: offering amount, sold/remaining, min investment, # investors, industry, revenue range, security types, exemptions, and the officers/directors/promoters.
`get_form_c_details(accession_or_url, cik=None)`	Parse a Form C (Reg CF) raise: target/max amount, price, security type, deadline, intermediary, employees, and a two-year financial snapshot (revenue, net income, assets, debt).
`get_company_facts(cik_or_query)`	Headline financials from a public company's XBRL facts: latest annual revenue, gross/operating income, net income, assets, liabilities, equity, cash.
`get_filing_text(url, offset=0, max_chars=20000)`	Fetch a document's text (HTML stripped) for reading/summarizing — paginated, since filings can exceed 1M characters.
`get_insiders(cik_or_query, limit=25)`	A company's insiders (officers, directors, >10% owners) from recent Section 16 filings, with roles.
`get_insider_trades(cik_or_query, limit=20)`	Recent insider transactions (Form 4): owner, role, buy/sell/grant, shares, price, shares owned after.

Install

Requires Python 3.12+. The zero-clone way to run it (the PyPI package is mcpwright-edgar; the command, server, and tools are all "edgar"):

uvx mcpwright-edgar

Claude Code

claude mcp add edgar -- uvx mcpwright-edgar

Claude Desktop

Add to claude_desktop_config.json:

{
  "mcpServers": {
    "edgar": { "command": "uvx", "args": ["mcpwright-edgar"] }
  }
}

OpenAI Agents SDK (Python)

It's a standard MCP server, so it works with any MCP-capable client — not just Claude. With the OpenAI Agents SDK:

from agents import Agent, Runner
from agents.mcp import MCPServerStdio

async def main():
    async with MCPServerStdio(
        name="edgar",
        params={
            "command": "uvx",
            "args": ["mcpwright-edgar"],
            "env": {"EDGAR_MCP_USER_AGENT": "your-app you@example.com"},
        },
    ) as edgar:
        agent = Agent(
            name="Analyst",
            instructions="Use the EDGAR tools for SEC filings and company data.",
            mcp_servers=[edgar],
        )
        result = await Runner.run(
            agent, "Recent Reg D raises in California — who's behind the biggest?"
        )
        print(result.final_output)

Any other MCP client (Cursor, VS Code, Cline, Goose, Zed, …)

They all launch a stdio MCP server the same way — point yours at:

{
  "mcpServers": {
    "edgar": {
      "command": "uvx",
      "args": ["mcpwright-edgar"],
      "env": { "EDGAR_MCP_USER_AGENT": "your-app you@example.com" }
    }
  }
}

Hosted chat connectors (e.g. ChatGPT connectors) expect a remote MCP server over Streamable HTTP; mcpwright-edgar runs locally over stdio. Running it behind Streamable HTTP for a hosted endpoint is straightforward if you need that.

SEC etiquette: the SEC requires a descriptive User-Agent with contact info and rate-limits to ~10 req/s. Set your own via the EDGAR_MCP_USER_AGENT env var (e.g. "your-app your-email@example.com"). The client throttles and retries for you.

Caching: responses are cached in-memory (byte-budgeted LRU) to cut latency and SEC load — immutable filing-archive content for days, the ticker map for 24h, everything else briefly. Set EDGAR_MCP_CACHE=0 to disable.

Develop

git clone https://github.com/mcpwright/edgar-mcp && cd edgar-mcp
uv sync
uv run pytest                       # tests (mocked SEC responses)
uv run ruff check . && uv run ruff format --check .   # lint + format
uv run mypy src tests               # strict type checking
uv run mcp dev src/edgar_mcp/server.py   # poke the tools in the MCP Inspector

Roadmap

Privacy

edgar-mcp runs entirely on your machine and collects, stores, or transmits no personal data — no accounts, no tracking, no telemetry. Its only outbound requests go to the U.S. SEC's EDGAR services (data.sec.gov, efts.sec.gov, www.sec.gov) to fetch the public filings you ask for; no API key is needed. One honest note: the SEC's fair-access policy asks for a descriptive User-Agent with contact info (EDGAR_MCP_USER_AGENT="your-app you@example.com") — whatever you set there is sent to the SEC with each request, and nowhere else. Responses are cached in memory only; nothing is persisted to disk.

Full policy: https://mcpwright.com/privacy/

Questions & feedback

Questions, ideas, or "could it do X?" → Discussions
Bugs & concrete feature requests → Issues

Contributions welcome — and if you build something with it, I'd love to hear about it.

Part of mcpwright · built by Devender Gollapally