# korean-dart-mcp [Health: Active]

**Category:** 💰 Finance & Fintech  
**Repository:** https://github.com/chrisryugj/korean-dart-mcp  
**GitHub Stars:** 119  
**npm Downloads (last month):** 2774  
**Views:** 0  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/korean-dart-mcp

## Description
OpenDART 전자공시MCP | 83개 API → 15 MCP 도구. 공시·재무·지분·XBRL + 버핏급 애널리스트 프레임(내부자 시그널·회계 리스크·퀄리티 체크리스트) + HWP/PDF 첨부 마크다운화 | 83 Korean DART disclosure APIs → 15 MCP tools with Buffett-grade analyst frames

## Claude Desktop Quick Installation
Install path detected from listing signals. Uses `npx` (confidence: high):

```json
"mcpServers": {
  "korean-dart-mcp": {
    "command": "npx",
    "args": ["-y","korean-dart-mcp"]
  }
}
```

## Documentation

## What korean-dart-mcp does

korean-dart-mcp exposes Korean OpenDART electronic disclosure data to MCP-compatible AI clients. The package consolidates 83 OpenDART API endpoints into 15 tools covering filings, company financials, ownership reports, capital events, XBRL statements, and document attachments.

Its analysis-oriented tools add structured views on top of raw disclosures. `insider_signal` groups reported insider buying and selling into cluster signals, while `disclosure_anomaly` scores selected accounting and governance indicators and returns supporting evidence. `buffett_quality_snapshot` compares multi-year quality measures such as ROE, debt-to-equity, revenue growth, and earnings growth against a checklist.

DART XML filings can be rendered with headings and tables, while attached HWP, HWPX, PDF, DOCX, and XLSX files can be extracted for searching and summarization.

## How it works

The server uses the OpenDART API as its disclosure source and runs as an MCP server for clients such as Claude Desktop, Cursor, Windsurf, Claude Code, VS Code, Gemini CLI, and Zed. A setup command interactively collects an OpenDART authentication key, detects supported installed clients, and updates their configuration.

Disclosure search includes preset queries for common event types, including treasury-share transactions, annual reports, capital increases, convertible bonds, mergers, and splits. Searches spanning more than 90 days without a company code are divided into shorter requests to work around the whole-market query limit. `get_corporate_event` can combine multiple event types into a timeline, and `get_major_holdings` combines major-shareholder and executive-shareholding data.

XBRL extraction can return a fuller Markdown representation of balance-sheet, income-statement, and cash-flow accounts. It parses presentation and calculation relationships and can check whether parent values reconcile with their component accounts, including industry-specific taxonomy prefixes such as the DX namespace used by financial groups.

## Setup and configuration

Node.js 20.19 or newer is required. The quickest installation path is:

```bash
npx -y korean-dart-mcp setup
```

The wizard asks for an OpenDART authentication key. The key can be skipped during setup and configured later. OpenDART provides the key through its service registration process.

The wizard then offers client selections for Claude Desktop, Cursor, Claude Code, Windsurf, VS Code, Gemini CLI, and Zed. It patches the selected configuration files and instructs the user to restart the client. On Windows, the generated configuration wraps `npx` with `cmd /c` to avoid command discovery problems.

## Tools and capabilities

Named capabilities include:

- `search_disclosures` for date-bounded filing searches and preset event filters.
- `download_document` for converting DART XML documents to Markdown.
- `get_attachments` for listing or extracting filing attachments.
- `get_xbrl` for structured XBRL accounts, taxonomy relationships, and calculation checks.
- `get_corporate_event` for capital-event timelines.
- `get_major_holdings` for 5% ownership and executive or major-shareholder reports.
- `insider_signal` for aggregated insider buying and selling activity.
- `disclosure_anomaly` for disclosure-based risk scoring and evidence.
- `buffett_quality_snapshot` for multi-year quality metrics and checklist comparisons.

## Limitations and notes

The package is intended for research assistance, not automated investment decisions. It focuses on Korean OpenDART disclosures and does not provide charting or real-time stock prices. Interpreting outputs requires familiarity with measures such as ROE, CAGR, and debt ratios. Large attachment extractions can produce substantial Markdown and may benefit from an AI client with a larger context window. An OpenDART authentication key is needed for normal data access.

_Full upstream README: https://allmcps.com/mcp/korean-dart-mcp/readme_