io.github.aion-org/edgar-answers-mcp
Provides normalized SEC EDGAR data (financials, filings, insider trades) to AI agents via MCP.
README
<!-- mcp-name: io.github.aion-autonomous-org/edgar-answers-mcp -->
EDGAR Answers — SEC filings for AI agents
Give your agent answers from SEC EDGAR, not raw API responses.
This repository holds the documentation and registry manifest for a hosted MCP server. There is nothing to install or run: connect to the endpoint below from any MCP client. The server itself is operated as a managed service.
Every US public company files its financials with the SEC, for free, in XBRL. The catch is that "free and public" is not the same as "usable": companies tag the same line item a dozen different ways, annual and quarterly figures live in the same stream, restatements silently overwrite history, and Form 4 insider filings arrive as raw XML. Existing EDGAR MCP servers hand your agent that mess and wish it luck.
This server does the parsing first. Your agent asks for revenue; it gets revenue.
Tools
get_financials — normalized annual statements
Multi-year income statement, balance sheet, and cash flow for any US-listed company, from a ticker or CIK.
get_financials(company: "AAPL", years: 4)
→ revenue, grossProfit, operatingIncome, netIncome, epsDiluted,
operatingCashFlow, capex, totalAssets, totalLiabilities,
stockholdersEquity, cashAndEquivalents, longTermDebt, sharesDiluted
— one clean row per fiscal year
Behind that: an ordered tag-fallback chain per line item (filers disagree on
which us-gaap tag means "revenue"), duration filtering so annual figures aren't
contaminated by quarterly cumulatives, and restatement resolution to the most
recently filed value. Unreported items come back null rather than guessed.
filing_diff — what changed year over year
filing_diff(company: "NVDA", section: "risk_factors")
→ paragraphs added and removed between the two most recent 10-Ks
The question an analyst actually asks — what did they start saying this year, and what did they quietly drop? — answered at paragraph level, with boilerplate that merely reflowed filtered out.
insider_activity — Form 4, parsed
insider_activity(company: "MSFT", filings: 10)
→ who, role, transaction type, shares, price, holdings after — tidy rows
Open-market buys separated from tax withholding and option exercises, so your agent doesn't mistake a scheduled vest for a conviction purchase.
filing_section — any section as clean text
filing_section(company: "TSLA", section: "risk_factors" | "mdna" | "business" | "1A")
Table-of-contents decoys are filtered out; you get the real section body.
search_filings — full-text search since 2001
search_filings(query: "\"supply chain disruption\"", forms: "10-K", startDate: "2025-01-01")
Connect
Streamable HTTP endpoint:
https://aion-org--edgar-answers-mcp.apify.actor/mcp
Authenticate with your Apify API token as a bearer token. In an MCP client config:
{
"mcpServers": {
"edgar-answers": {
"url": "https://aion-org--edgar-answers-mcp.apify.actor/mcp",
"headers": { "Authorization": "Bearer YOUR_APIFY_TOKEN" }
}
}
}
For clients that only speak stdio, bridge with npx mcp-remote <url> --header "Authorization: Bearer YOUR_APIFY_TOKEN".
The server runs in standby mode: it scales to zero when idle and wakes on your first request, so the first call after a quiet period takes a few seconds.
Pricing
Pay per successful call — failed calls are not billed.
| Call | Price |
|---|---|
search_filings |
$0.005 |
get_financials |
$0.02 |
insider_activity |
$0.02 |
filing_section |
$0.03 |
filing_diff |
$0.05 |
Data and compliance
Source data is SEC EDGAR — US government filings, public domain. Requests are made with a declared User-Agent and rate-limited well under the SEC's published ceiling, per their access guidelines. Nothing here is scraped from a site that prohibits it.
Limits, stated plainly
- US GAAP filers. Foreign private issuers reporting under IFRS return no
normalized financials;
get_financialssays so rather than guessing. - Coverage follows EDGAR. Full-text search reaches back to 2001; XBRL financials exist from roughly 2009 onward.
- Not investment advice. This is filing data, faithfully parsed. What you conclude from it is yours.
filing_diffcompares the two most recent filings of a form; arbitrary filing pairs aren't exposed yet.
Found a parsing bug or want a tool that isn't here? Open an issue here or on the Actor — specific reports about specific tickers get fixed.
Listed on the Apify Store. Documentation in this repository is © Aion; the service is proprietary.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.
E2B
Using MCP to run code via e2b.