MCP Video Generation with Veo2

What is MCP Video Generation with Veo2?

MCP Video Generation with Veo2 is an MCP server that exposes Google’s Veo2 video generation model to clients. It lets users create videos from text prompts or images and access the generated videos via MCP resources. Intended for developers and AI practitioners who want to integrate AI video generation into their workflows.

How to use MCP Video Generation with Veo2?

Install Node.js 18+ and obtain a Google API key with Veo2/Gemini access. Clone the repo, install dependencies (npm install), create a .env file with your GOOGLE_API_KEY, then build (npm run build). Start the server with npm start for stdio transport or npm start sse for SSE (port 3000). Use the exposed MCP tools generateVideoFromText, generateVideoFromImage, and listGeneratedVideos.

Key features of MCP Video Generation with Veo2

• Generate videos from text prompts
• Generate videos from images
• Access generated videos through MCP resources
• Supports both stdio and SSE transports
• Configurable resolution and duration (5-8 seconds)

Use cases of MCP Video Generation with Veo2

• Automatically generate short video clips from textual descriptions
• Create videos from user-uploaded images or AI-generated images
• Build AI‑powered video creation tools for content marketing
• Integrate video generation into MCP‑based agents or chatbots

FAQ from MCP Video Generation with Veo2

What API key is required?

A Google API key with access to the Gemini API and Veo2 model. You must set up a credit card with your API key via aistudio.google.com.

What are the system requirements?

Node.js 18 or higher. No other runtime dependencies are required.

Where are generated videos stored?

By default in ./generated-videos. You can change the storage directory by setting the STORAGE_DIR environment variable.

What configuration options does video generation support?

You can set aspect ratio ("16:9" or "9:16"), person generation ("dont_allow" or "allow_adult"), number of videos (1 or 2), duration (5–8 seconds), prompt enhancement, and a negative prompt. When using image input, personGeneration only supports "dont_allow".

Which transports are supported?

Both stdio (default) and SSE transports are supported. For SSE, the server listens on the port specified in the PORT environment variable (default 3000).