An MCP server to use Sora video generation APIs
Configuration
View docs{
"mcpServers": {
"doriandarko-sora-mcp": {
"url": "http://localhost:3000/mcp",
"headers": {
"DOWNLOAD_DIR": "/Users/you/Downloads",
"OPENAI_API_KEY": "sk-..."
}
}
}
}You run a Model Context Protocol (MCP) server that integrates with OpenAI’s Sora 2 API to generate and remix videos from prompts. This server enables local or remote clients to request video creation, track progress, and manage outputs, all through simple MCP-compatible commands and endpoints.
Use the two available transport modes to connect MCP clients either locally via stdio or remotely via HTTP. In stdio mode, Claude Desktop communicates directly with the local server process. In HTTP mode, web-based tools and MCP clients connect over the network to perform video creation, status checks, and downloads.
Prerequisites you need before installation:
- Node.js 18 or newer
- OpenAI API key with Sora access
- An MCP-compatible client (Claude, Cursor, VS Code, etc.)
Step-by-step installation commands you should run in your project directory:
# 1) Clone the project
git clone https://github.com/Doriandarko/sora-mcp
cd sora-mcp
# 2) Install dependencies
npm install
# 3) Build the project
npm run buildTwo MCP server implementations are provided to cover different use cases. Each server is designed for a specific transport and client type, ensuring optimal performance and security for its communication method.
Generate a video from a text prompt. Parameters include prompt, model, seconds, size, and optional input_reference.
Check the status and progress of a video generation job by providing the video_id.
List all video generation jobs with optional pagination and sort order.
Return a curl command to download a completed video, with options for format/variant.
Auto-download and save a completed video to a specified directory with an optional filename.
Create a remix of an existing video using a new prompt.
Delete a video job and its associated assets.