Dy

dy-mcp is a powerful automation tool for Douyin that helps you quickly leverage automated Douyin workflows. It enables you to parse share links to obtain watermark-free download URLs, upload videos from a specified path, and generate Drafts in editing apps. This guide shows you how to install, configure, and run the dy-mcp server so you can automate common Douyin tasks with an MCP client.

How to use

You will run the local MCP server and connect to it with an MCP client. The server exposes standard input/output (stdio) endpoints that allow you to start and manage tasks from your local environment. Use the provided stdio configuration to launch the server and then issue commands through your MCP client to perform actions such as parsing Douyin links, uploading media, and creating Drafts.

To operate, you configure a local runtime command that starts the server module. In this setup, the runtime uses your Python environment to execute server.py. You will then reference this server in your MCP client configuration so that your client can send requests to the local process and receive results.

How to install

Prerequisites you need before starting:

• Git is installed on your system
• Conda is available or you have a compatible Python environment
• Python 3.10 is used in the virtual environment
• Playwright is installed for browser automation

Step-by-step commands you should run in order:

# 1. Clone the project
git clone https://github.com/vipcong816/dy-mcp.git
cd dy-mcp

# 2. Create and activate a Python virtual environment (recommended)
conda create -n dy-mcp python=3.10
conda activate dy-mcp

# 3. Install dependencies
pip install -r requirements.txt

Additional configuration and runtime notes

Configuration is done inside the server script and in the runtime setup for MCP. You need to specify the local Chrome path used by the browser automation tasks and create folders to store cookies and uploaded files.

# 4. Start the Playwright browsers (at least chromium is required)
playwright install chromium

# 5. Create necessary folders in the project root
# - cookiesFile
# - videoFile

Start the server and MCP wiring

To run the server locally, use the Python script that serves the streamable HTTP interface. This enables you to launch the local environment and then connect via MCP.

python server_streamable_http.py

MCP stdio configuration

Configure the MCP client to launch the local Python process that runs the server. Use the following stdio configuration snippet.

{
  "mcpServers": {
    "dy_mcp": {
      "command": "C:/Users/你的用户名/miniconda3/envs/环境名/python.exe",
      "args": ["server.py"]
    }
  }
}

MCP Inspector configuration

To inspect and interact with the MCP server, you can run the inspector tool which helps validate the MCP setup and the available endpoints.

npx @modelcontextprotocol/inspector

Security and best practices

Keep your cookies and uploaded media in dedicated folders to avoid mixing with other project assets. Ensure the path to your local Chrome is correct in server.py to prevent automation errors. Regularly update dependencies and Browser drivers for compatibility.

Available tools

parse_dy_link

Parse Douyin share links to obtain a watermark-free download URL for the media.

upload_video

Upload a video from a specified path to the target platform.

generate_draft

Generate a Draft in the connected editing app to start post-production.