mcp-vsepgt-server

Here will be a set of MCP servers for VseGPT, written in Python. MCP (Model Context Protocol) is a protocol for interaction between language models and external context servers. The main tasks of MCP include: - Providing models access to up-to-date data and knowledge via API - The ability to perform actions in the real world (sending emails, working with files, etc.) - Ensuring security when working with external systems MCP servers operate as an intermediary layer between language models and external services/data, providing a standardized interface for interaction. ## Installation Python + [fastmcp package](https://github.com/jlowin/fastmcp) is used for operation (at the time of launch - version 0.4.1) ## Logic Different functions of VseGPT will usually be represented in different servers, not in one. This will allow dynamically enabling and disabling functionality provided to models in the tools field, without overloading this field, the model context, and confusing it. ## Servers ### mcp_gen_image.py Server for generating images through VseGPT. Files are saved in the script folder / tmp_images. It is launched as: `fastmcp run mcp_gen_image.py` Environment parameters: - VSEGPT_API_KEY=your_vsegpt_key (required) - IMG_MODEL_ID=ID of the image generation model on the VseGPT website (optional, defaults to `img-dummy/image`) - IMG_SIZE=size of generation (optional, defaults to `1024x1024`) An example of the setup can be found here: https://vsegpt.ru/ExtTools/CherryStudio ### mcp_gen_tts.py **(Beta, configure yourself through the code if necessary)** Server for generating speech through VseGPT. Files are saved in the script folder / tmp_images. Upon receiving, MPC-HC is launched to play the sound. It is launched as: `fastmcp run mcp_gen_tts.py` Environment parameters: - VSEGPT_API_KEY=your_vsegpt_key (required)