MiniMax MCP

Overview

MiniMax MCP is the official Python Model Context Protocol server for MiniMax Text to Speech, voice cloning, video generation, image generation, music generation, and voice design APIs. It is intended for MCP clients such as Claude Desktop, Cursor, Windsurf, OpenAI Agents, and similar tools.

Installation

Install uv/uvx first, for example curl -LsSf https://astral.sh/uv/install.sh | sh. The published package is minimax-mcp on PyPI and the documented MCP client transport runs it with uvx.

Configuration

The documented stdio transport uses command uvx with args minimax-mcp and -y. Configure MINIMAX_API_KEY, MINIMAX_API_HOST, optional MINIMAX_MCP_BASE_PATH, optional MINIMAX_API_RESOURCE_MODE, and optional FASTMCP_LOG_LEVEL. MINIMAX_API_HOST must match the API key region: Global uses https://api.minimax.io; Mainland uses https://api.minimaxi.com.

Transport

The README says the server supports stdio and SSE. The concrete package target submitted here is the documented/imported stdio transport. README describes SSE as suitable for local or cloud deployment, with URL resources recommended for cloud deployments, but does not provide a concrete SSE command, port, or hosted endpoint.

Capabilities

Tools include text_to_audio, list_voices, voice_clone, play_audio, generate_video, query_video_generation, text_to_image, music_generation, and voice_design.

Limitations

MiniMax API calls may incur costs. The API key and host must be regionally aligned. Local file handling depends on MINIMAX_MCP_BASE_PATH; otherwise local file paths must be absolute. For cloud/SSE usage, URL resources are recommended. Windows Claude Desktop users need Developer Mode enabled.