MCP Video Generation Server

A powerful MCP (Model Context Protocol) server for AI video generation using the latest 2025 models. Generate videos from text prompts or images using cutting-edge AI models like Google Veo-3, Tencent HunyuanVideo, Mochi-1, and many more.

Features

• 25+ AI Video Models: Latest 2025 models including Veo-3, HunyuanVideo, Mochi-1, LTX-Video, Pyramid Flow, Seedance, Hailuo, Kling, and more
• Multiple Input Types: Text-to-video, image-to-video, and hybrid models
• Advanced Parameters: Duration, resolution, FPS, aspect ratio, motion intensity, camera movement
• Video Processing Tools: Metadata extraction, speed adjustment, scaling, filtering, concatenation, audio separation
• Automatic Downloads: Generated videos are automatically saved locally with organized naming
• MCP Integration: Works with Claude Desktop, Cursor, Claude Code, and other MCP clients

Quick Start

Prerequisites

• Node.js 18+
• pnpm package manager
• Replicate API token
• FFmpeg (for video processing tools)

Installation

1. Clone and install dependencies:

git clone <repository-url>
cd mcp-video-generation
pnpm install

2. Set up environment variables:

cp .env.example .env.local
# Add your REPLICATE_API_TOKEN to .env.local

3. Build and install to MCP clients:

pnpm run install-server  # Install to all MCP clients
# Or install to specific clients:
pnpm run install-desktop # Claude Desktop only
pnpm run install-cursor  # Cursor only
pnpm run install-code    # Claude Code only

Supported Models

Google Veo Models (2025)

• veo-3: Latest model with audio generation, up to 30s, 4K support
• veo-3-fast: Fast variant with audio generation
• veo-2: Previous generation, text-to-video only

Tencent HunyuanVideo (2025)

• hunyuan-video: High-quality Chinese model, up to 10s, 720p-1280p

Genmo Mochi-1 (2025)

• mochi-1: Open-source model, 5.4s max, 480p

LTX Video (2025)

• ltx-video: Lightricks model, up to 10s, 768p

Pyramid Flow (2025)

• pyramid-flow: Both text-to-video and image-to-video

ByteDance Seedance (2025)

• seedance-pro: Professional model, both input types
• seedance-lite: Lightweight variant

MiniMax Hailuo (2025)

• hailuo-2: Both text and image inputs
• minimax-video: Text-to-video only
• minimax-director: Director model with camera controls

Kuaishou Kling (2025)

• kling-v2.1-master: Latest master model
• kling-v1.6-pro: Professional variant

Luma Ray (2025)

• ray-flash-2: Fast variant
• ray-2: Standard model
• luma-ray: Original Ray model

Alibaba WAN (2025)

• wan-t2v-720p/480p: Text-to-video variants
• wan-i2v-720p/480p: Image-to-video variants

PixVerse (2025)

• pixverse-v4.5: Latest version, up to 8s, multiple resolutions

Usage Examples

Basic Video Generation

// Generate a video with Veo-3
{
  "model": "veo-3",
  "prompt": "A serene sunrise over mountains with birds flying",
  "duration": 5,
  "resolution": "1080p",
  "aspect_ratio": "16:9",
  "generate_audio": true
}

Image-to-Video

// Convert image to video with Pyramid Flow
{
  "model": "pyramid-flow",
  "prompt": "The scene comes to life with gentle movement",
  "image": "https://example.com/image.jpg",
  "duration": 5,
  "resolution": "768p"
}

Director Mode with Camera Controls

// Use MiniMax Director for camera movement
{
  "model": "minimax-director",
  "prompt": "A cinematic shot of a bustling city street",
  "duration": 6,
  "camera_movement": "dolly",
  "resolution": "1080p"
}

Video Processing Tools

Metadata Extraction

// Get comprehensive video metadata
{
  "tool": "get-video-metadata",
  "file_path": "/absolute/path/to/video.mp4"
}

Speed Adjustment

// Change video playback speed
{
  "tool": "adjust-video-speed",
  "input": "/path/to/input.mp4",
  "speed_factor": 2.0,  // 2x speed
  "output": "/path/to/output.mp4"
}

Video Scaling

// Resize video resolution
{
  "tool": "scale-video",
  "input": "/path/to/input.mp4",
  "width": 1920,
  "height": 1080,
  "output": "/path/to/output.mp4",
  "maintain_aspect": true
}

Apply Filters

// Apply FFmpeg filters
{
  "tool": "apply-video-filters",
  "input": "/path/to/input.mp4",
  "filter_string": "brightness=0.1,contrast=1.2",
  "output": "/path/to/output.mp4"
}

Concatenate Videos

// Join multiple videos
{
  "tool": "concatenate-segments",
  "inputs": ["/path/to/video1.mp4", "/path/to/video2.mp4"],
  "output": "/path/to/combined.mp4",
  "re_encode": false
}

Separate Audio and Video

// Extract audio and video tracks
{
  "tool": "separate-audio-and-video",
  "input": "/path/to/input.mp4",
  "video_output": "/path/to/video_only.mp4",
  "audio_output": "/path/to/audio_only.mp3",
  "audio_format": "mp3"
}

Configuration

Environment Variables

Create .env.local file with:

REPLICATE_API_TOKEN=your_replicate_token_here

MCP Client Configuration

The server automatically updates configuration files for:

• Claude Desktop: ~/Library/Application Support/Claude/claude_desktop_config.json
• Cursor: ~/Library/Application Support/Cursor/User/claude_desktop_config.json
• Claude Code: ~/.claude/claude_code_config.json
• MCP: .mcp.json

Development

Commands

pnpm run build        # Compile TypeScript
pnpm start            # Run the server
pnpm run install-server # Build and install to all clients

Project Structure

src/
├── index.ts          # Main MCP server implementation
scripts/
├── update-config.js  # Configuration installer
dist/                 # Compiled JavaScript
videos/               # Default video output directory

Adding New Models

1. Add model configuration to VIDEO_MODELS object in src/index.ts
2. Add model to the enum in the tool schema
3. Add model-specific parameter handling in the switch statement
4. Build and reinstall: pnpm run install-server

Troubleshooting

Common Issues

FFmpeg not found

# macOS
brew install ffmpeg

# Ubuntu/Debian
sudo apt install ffmpeg

Replicate API errors

• Verify your API token is correct
• Check your Replicate account credits
• Ensure the model version is still available

Video generation timeout

• Large videos may take up to 15 minutes
• Monitor progress using the web URL provided
• Check Replicate dashboard for queue status

File permissions

• Ensure output directories are writable
• Check file paths are absolute and valid

License

MIT

Contributing

1. Fork the repository
2. Create a feature branch
3. Make your changes
4. Add tests if applicable
5. Submit a pull request

Support

For issues and questions:

• Check the troubleshooting section
• Review Replicate model documentation
• Open an issue on GitHub