vedit-mcp

Vedit-MCP

This is an MCP service for video editing, which can achieve basic editing operations with just one sentence.

English | 中文

Quick Start

1. Install Dependencies

1.1 Clone this project or directly download the zip package

1.2 Configure the Python environment

1. It is recommended to use uv for installation

cd vedit-mcp
uv pip install -r requirements.txt

2. Or install directly using pip

pip install -r requirements.txt

1.3 Configure ffmpeg

vedit-mcp.py relies on ffmpeg for implementation. Therefore, please configure ffmpeg.

# For Mac
brew install ffmpeg
# For Ubuntu
sudo apt update
sudo apt install ffmpeg

2. Start the Service

2.1. It is recommended to use google-adk to build your own project

• Please refer to adk-sample

Before executing this sample script

1. Please ensure that the path format is at least as follows

• sample
  • kb
    • raw/test.mp4 // This is the original video you need to process
  • adk_sample.py
• vedit_mcp.py

2. Please install the following two dependencies

# # adk-sample pip install requirements
# google-adk==0.3.0
# litellm==1.67.2

3. Please set the api-key and api-base

Currently, this script uses the API of the Volcano Ark Platform, and you can go there to configure it by yourself.

After obtaining the API_KEY, please configure the API_KEY as an environment variable.

export OPENAI_API_KEY="your-api-key"

4. Execute the script

cd sample
python adk_sample.py

5. End of execution

After this script is executed correctly and ends, a video result file will be generated in kb/result, and a log file will be generated and the result will be output.

If you need secondary development, you can choose to add vedit_mcp.py to your project for use.

2.2 Or build using cline

Firstly, please ensure that your Python environment and ffmpeg configuration are correct Configure cline_mcp_settings. json as follows

{
  "mcpServers": {
    "vedit-mcp": {
      "command": "python",
      "args": [
        "vedit_mcp.py",
        "--kb_dir",
        "your-kb-dir-here"
      ]
    }
  }
}

2.3. Execute using the stramlit web interface

To be supplemented

3. precautions

1. It is recommended to use the thinking model to handle this type of task. Currently, it seems that the thinking model performs better in handling this type of task? But no further testing has been conducted, it's just an intuitive feeling.

Shared threads will appear here, showcasing real-world applications and insights from the community. Check back soon for updates!