Upstage MCP Server

🚀 Upstage Document Processing Server (MCP Server)

The Upstage Document Processing Server (MCP Server) is a tool for parsing and extracting document information. It supports multiple file formats and can extract information based on predefined or automatically generated patterns. This server aims to help AI models process and understand document content more efficiently.

🚀 Quick Start

To start using the Upstage Document Processing Server, you need to meet the following prerequisites:

1. Obtain a valid Upstage API key by visiting Upstage Console.
2. Install a Python environment and ensure all dependencies are installed (specific requirements can be found in the project documentation).

✨ Features

Document Parsing (parse_document)

• Description: Process the document and extract its content while preserving the structure.
• Parameters: file_path - The path of the document to be parsed.
• Example Query: Please parse the document located at C:\Users\username\Documents\contract.pdf and provide a summary.

Information Extraction (extract_information)

• Description: Extract structured information from the document based on predefined or automatically generated patterns.
• Parameters: file_path - The document path; schema_path (optional) - A JSON file containing the extraction pattern; auto_generate_schema (default is true) - Whether to automatically generate the pattern.
• Example Query: Please extract the invoice number, date, and total amount from C:\Users\username\Documents\invoice.pdf.

📦 Installation

Using the uv Tool

1. Install uvx and add it to the system path.
2. Start the server by running the following command in the terminal:

uvx run --app upstage-mcp-server

Configuring Claude Desktop

Edit the claude_desktop_config.json file and add the following configuration:

{
  "mcp_server": {
    "upstage_mcp_server": {
      "executable_path": "path_to_your_uv_executable",
      "args": [
        "run",
        "--directory",
        "path_to_upstage_mcp_server_directory",
        "python",
        "-m",
        "upstage_mcp_server.server"
      ],
      "env": {
        "UPSTAGE_API_KEY": "your_api_key_here"
      }
    }
  }
}

Output Directory

After the server processes the document, the output results are located at:

• Windows: %APPDATA%\Claude\outputs
• macOS: ~/Library/Logs/Claude/output

Local Development Setup

Installing Dependencies

Install the necessary Python packages using the following command:

pip install upstage-mcp-server

Running the Server

Run the following command in the project root directory:

python -m upstage_mcp_server.server

💻 Usage Examples

Basic Usage

The following are the basic usage methods of the two main functions:

Document Parsing

# Assume the function call is made in Python
# Replace 'your_file_path' with the actual file path
result = parse_document(file_path='your_file_path')

Information Extraction

# Assume the function call is made in Python
# Replace 'your_file_path' with the actual file path
result = extract_information(file_path='your_file_path', schema_path='your_schema_path', auto_generate_schema=True)

Troubleshooting

Common Issues

1. Missing API Key
   • Ensure that UPSTAGE_API_KEY is correctly set in the claude_desktop_config.json file.
2. File Not Found
   • Check the absoluteness of the file path and ensure that no special characters are not properly escaped.
3. Server Startup Failure
   • Confirm that the virtual environment is activated and all dependencies are installed. Check the Claude Desktop log file.

Log Files

• Windows: %APPDATA%\Claude\logs
• macOS: ~/Library/Logs/Claude

Contribution Guidelines

Community contributions are welcome! Please refer to the contribution guidelines in the project documentation to learn how to participate in development and submit code.

📄 License

This software is licensed under the MIT License.