SEC EDGAR MCP Server: Filings, XBRL Financials and Insider Trades for AI Agents

Give Claude, Cursor, ChatGPT, VS Code Copilot or your own AI agent live access to SEC EDGAR, the official US database of public company filings. Your agent can pull financial statements for any US-listed company, search every 10-K, 10-Q and 8-K since 2001 by keyword, read Risk Factors or MD&A as clean text, rank companies on any XBRL metric, and track insider buying and selling (Form 4).

No SEC API key, no scraping and no parsing on your side. The server returns compact JSON that LLMs handle well, and you pay only for tool calls that succeed.

• Official data: comes straight from data.sec.gov and efts.sec.gov (SEC EDGAR APIs). US government data is public domain.
• 8 tools built for agents: clear descriptions, typed arguments, structured output and helpful error messages.
• Fast: responses are cached, requests are throttled to the SEC fair-access limit and retried automatically.
• Pay per tool call: from $0.01 per call, with no subscription and no monthly minimum.

What can you ask it?

• "Show Nvidia's last 8 quarters of revenue, gross margin and free cash flow."
• "Which 10-K filings from 2026 mention 'going concern' and 'tariffs'? Group them by industry."
• "Summarize the new risk factors in Tesla's latest 10-K compared with the previous one."
• "Rank the top 25 US companies by net income in CY2025."
• "Have Apple insiders bought or sold stock in the open market recently?"
• "Compare R&D spend of AMD, Intel and Nvidia over 5 years."
• "What did Microsoft disclose in its latest 8-K?"

Tools

Tool	What it does	Price
search_companies	Finds a company by name or ticker and returns its CIK, ticker and exchange	$0.01
get_company_filings	Returns the company profile (SIC industry, fiscal year end, address, former names) and its filings, filtered by form type and date, with direct document URLs	$0.01
search_filings_full_text	Full-text search across all EDGAR filings and exhibits since 2001. Supports exact phrases, OR and exclusions, plus form, date and company filters	$0.01
get_financial_statements	Revenue, gross profit, operating income, net income, EPS, operating cash flow, capex, free cash flow, cash, assets, liabilities, debt, equity and margins, annual or quarterly	$0.01
get_xbrl_concept	Full time series of any XBRL line item (for example InventoryNet or NumberOfEmployees). Leave out concept to discover which concepts the company reports	$0.01
compare_companies_metric	Ranks all ~6,000 reporting companies, or your own peer list, on one metric for a period. Aliases such as revenue or net_income merge the different XBRL tags that companies use	$0.01
read_filing	Reads any filing as clean text, either whole or one section: business, risk_factors, mdna, financial_statements, legal_proceedings, cybersecurity, market_risk, controls, properties, executive_compensation. Long documents are paged	$0.02
get_insider_transactions	Parses recent Form 4 filings: insider name and role, buy or sell, shares, price, value, holdings after the trade and the 10b5-1 flag, plus a buy/sell summary	$0.02

Every tool accepts a ticker (AAPL), a CIK (320193) or a company name (Apple).

How to connect (MCP)

The server uses the MCP Streamable HTTP transport at:

https://gazidev--sec-edgar-mcp.apify.actor/mcp

Authenticate with your Apify API token, which you can find under Apify Console > Settings > API & Integrations:

Authorization: Bearer YOUR_APIFY_TOKEN

Claude Code

claude mcp add --transport http sec-edgar https://gazidev--sec-edgar-mcp.apify.actor/mcp \
  --header "Authorization: Bearer YOUR_APIFY_TOKEN"

Claude Desktop

Edit claude_desktop_config.json:

{
  "mcpServers": {
    "sec-edgar": {
      "command": "npx",
      "args": [
        "-y", "mcp-remote",
        "https://gazidev--sec-edgar-mcp.apify.actor/mcp",
        "--header", "Authorization: Bearer YOUR_APIFY_TOKEN"
      ]
    }
  }
}

Cursor

Edit ~/.cursor/mcp.json, or .cursor/mcp.json in your project:

{
  "mcpServers": {
    "sec-edgar": {
      "url": "https://gazidev--sec-edgar-mcp.apify.actor/mcp",
      "headers": { "Authorization": "Bearer YOUR_APIFY_TOKEN" }
    }
  }
}

VS Code (GitHub Copilot agent mode)

Edit .vscode/mcp.json:

{
  "servers": {
    "sec-edgar": {
      "type": "http",
      "url": "https://gazidev--sec-edgar-mcp.apify.actor/mcp",
      "headers": { "Authorization": "Bearer YOUR_APIFY_TOKEN" }
    }
  }
}

Windsurf, Cline, Zed and other MCP clients

Add a remote or HTTP MCP server with the URL above and the Authorization header. For clients that only support stdio, wrap the server with npx mcp-remote <url> --header "Authorization: Bearer <token>".

Python (FastMCP client)

import asyncio
from fastmcp import Client
from fastmcp.client.transports import StreamableHttpTransport

transport = StreamableHttpTransport(
    "https://gazidev--sec-edgar-mcp.apify.actor/mcp",
    headers={"Authorization": "Bearer YOUR_APIFY_TOKEN"},
)

async def main():
    async with Client(transport) as client:
        result = await client.call_tool("get_financial_statements", {"company": "NVDA", "period": "quarterly"})
        print(result.data)

asyncio.run(main())

Batch mode (no MCP client needed)

You can also start the Actor as a normal run from the Apify Console, the API, a schedule or an integration such as Make, Zapier or n8n. Pick a tool, give its arguments as JSON, and the results go to the dataset:

{
  "tool": "get_financial_statements",
  "arguments": { "company": "MSFT", "period": "annual", "limit": 5 }
}

To run several calls at once, use calls:

{
  "calls": [
    { "tool": "get_insider_transactions", "arguments": { "company": "AAPL", "open_market_only": true } },
    { "tool": "search_filings_full_text", "arguments": { "query": "\"material weakness\"", "form_types": ["10-K"], "date_from": "2026-01-01" } }
  ]
}

Batch runs cost the same per tool call as MCP calls.

Example output (get_financial_statements, trimmed)

{
  "company": { "cik": 320193, "name": "Apple Inc.", "ticker": "AAPL" },
  "period": "annual",
  "periods": [
    {
      "period_end": "2025-09-27",
      "revenue": 416161000000,
      "gross_profit": 195201000000,
      "operating_income": 133050000000,
      "net_income": 112010000000,
      "eps_diluted": 7.46,
      "operating_cash_flow": 111482000000,
      "capital_expenditures": 12715000000,
      "free_cash_flow": 98767000000,
      "total_assets": 359241000000,
      "stockholders_equity": 73733000000,
      "gross_margin_pct": 46.91,
      "net_margin_pct": 26.92,
      "source_form": "10-K",
      "source_accession": "0000320193-25-000079"
    }
  ]
}

Pricing

This Actor uses pay-per-event pricing. You pay only for tool calls that succeed:

Event	Price
Data tool call (search, filings, full-text search, financials, XBRL concept, compare)	$0.01
Document tool call (read_filing, get_insider_transactions)	$0.02
Actor start (once per Standby session)	$0.005

• Failed calls are free. A call that errors (for example, an unknown ticker) is not charged.
• A typical research session of 10 tool calls costs about $0.10 to $0.15.
• Apify's free plan includes monthly credits, so you can try the server at no cost.
• You can set a maximum charge per run in Apify. When the limit is reached, the tools return a clear message instead of charging more.

FAQ

Is this the official SEC API? The data comes from the SEC's official public EDGAR APIs. This Actor is an independent project and is not affiliated with or endorsed by the U.S. Securities and Exchange Commission.

Do I need an SEC API key? No. EDGAR does not require keys. The server handles the SEC's User-Agent and rate-limit requirements (10 requests/second) for you.

How fresh is the data? The data is live. New filings appear in EDGAR within minutes of acceptance. The server caches responses briefly: 10 minutes for filing lists, 1 hour for XBRL facts and 15 minutes for full-text search.

Which companies are covered? All SEC filers. Name and ticker search covers the roughly 10,000 companies with an exchange ticker. For any other filer, including private companies that file Form D, use the CIK. XBRL financials cover filings since about 2009. Full-text search covers filings since 2001.

Can it handle foreign companies (20-F filers)? Yes. get_company_filings, read_filing and full-text search work for all filers. Companies that report under IFRS, such as TSMC, have no US-GAAP statements, so use get_xbrl_concept with taxonomy: "ifrs-full". The tool tells you when this applies.

Why is a metric missing for a company? Companies choose which XBRL elements to tag. get_financial_statements already falls back through common alternatives, for example Revenues, then RevenueFromContractWithCustomerExcludingAssessedTax, then SalesRevenueNet. For anything else, call get_xbrl_concept without a concept to list what the company reports.

Are quarterly numbers accurate for Q4 and cash flow? 10-Qs report cash flow year-to-date, and most companies never tag Q4 on its own. The server derives these quarters from year-to-date differences, for example FY minus 9 months, and lists them in derived_fields. Derived EPS is approximate.

Can I use the data commercially? Yes. SEC EDGAR content is US government work and is not subject to copyright. Filings may contain third-party material, such as exhibits, so check with your own counsel for edge cases.

Is this investment advice? No. The server provides raw regulatory data. Always verify important figures against the original filing; every result includes source URLs or accession numbers.

Something is broken or missing? Open an issue on the Actor's Issues tab. Feature requests are welcome, for example 13F holdings or XBRL segment data.

Keywords

SEC EDGAR API, SEC filings MCP, 10-K API, 10-Q, 8-K, XBRL financial statements API, company fundamentals API, EDGAR full-text search, insider trading Form 4 API, risk factors extraction, MD&A, financial data for LLMs, MCP server for finance, Claude MCP finance, Cursor MCP, AI agent stock research.