home / mcp / opendart mcp server

OpenDART MCP Server

Provides structured access to OpenDART data via MCP, including disclosures, reports, ownership changes, dividends, and risk signals.

Installation
Add the following to your MCP client configuration file.

Configuration

View docs
{
  "mcpServers": {
    "changoolee-mcp-opendart": {
      "command": "YOUR_LOCATION/.venv/bin/mcp-opendart",
      "args": [],
      "env": {
        "HOST": "0.0.0.0",
        "PORT": "8000",
        "LOG_LEVEL": "INFO",
        "TRANSPORT": "stdio",
        "MCP_SERVER_NAME": "opendart_mcp",
        "OPENDART_API_KEY": "API-KEY",
        "OPENDART_BASE_URL": "https://opendart.fss.or.kr/api/"
      }
    }
  }
}

You can run MCP OpenDART to securely interact with OpenDART data through a contextual AI assistant. This server lets you query disclosures, financial reports, ownership changes, and other corporate information while preserving data privacy and enabling AI-driven insights.

How to use

Connect your MCP client to the OpenDART MCP server and start asking questions that require regulatory disclosures, financial statements, and corporate events. You can request quarterly or annual reports, search for major disclosures, analyze financial metrics, and track real-time updates on important filings. The server handles data access, parsing of financial notes, and structured extraction so your AI assistant can build context, summarize findings, and surface risk signals. Use it to get authoritative figures on salaries, ownership changes, dividends, and potential liabilities, then combine those insights with AI-driven analysis for investment or research purposes.

How to install

Prerequisites: Python 3.10 or newer installed on your system. You also need Git to clone the repository and a working network to install dependencies.

# 1) Clone the MCP OpenDART project
git clone https://github.com/ChangooLee/mcp-opendart.git
cd mcp-opendart

# 2) Create and activate a Python virtual environment (Python 3.10+ required)
python3.10 -m venv .venv
source .venv/bin/activate

# 3) Install dependencies in editable mode
python -m pip install --upgrade pip
pip install -e .

Additional sections

Configuration and startup use the MCP server configuration shown in the guidance. You will run a local MCP server instance that exposes an endpoint you can connect to from your MCP client. The server requires an API key for OpenDART access and supports both stdio (local) and HTTP transport modes depending on your setup.

Security and access control: keep your API key secret, store sensitive values in environment variables, and enable appropriate logging levels for troubleshooting. Use a proper host and port in production and consider enabling rate limiting and network-level protections. If you are using Docker, you can run the container with environment variables to configure the server without embedding secrets in code.

Troubleshooting tips: ensure Python 3.10+ is active in the shell where you run the server, verify that the container or process has network access to OpenDART, and check logs for authentication failures or transport errors. If you encounter data access delays, confirm required permissions and any additional access controls on certain data sources.

Available tools

ds001_disclosure

Search and retrieve corporate disclosures and related documents from the OpenDART dataset.

ds002_periodic

Query periodic financial reports and extract key information such as executives, dividends, and ownership notes.

ds003_financial

Access and analyze financial statements, notes, and related financial data across filings.

ds004_ownership

Analyze ownership structures and changes among major shareholders and related parties.

ds005_major

Retrieve major reports and executive summaries to understand material business developments.

ds006_securities

Fetch and analyze securities filings for corporate events, risk factors, and warrants.