Submagic MCP Server

mcp-name: io.github.sidart10/submagic-mcp-server

Python MCP server for AI-powered video editing using the Submagic API. Supports automatic captions in 107 languages, magic zooms, B-rolls, silence removal, and viral clip generation from YouTube videos.

Features

• Automatic AI captions with 98%+ accuracy in 107 languages
• Magic zooms (AI-powered dynamic emphasis effects)
• Magic B-rolls (auto-insert stock footage with configurable coverage)
• Silence removal with three speed presets
• Filler word removal (removes "um", "uh", "like")
• Custom B-roll insertion at exact timestamps
• YouTube to short-form clips (TikTok, Reels, Shorts)
• 30+ professional templates (Hormozi, Beast, Sara)
• Platform-optimized exports (9:16, 1:1, 16:9, 4:5)
• Post-creation editing and iterative refinement

Installation

From PyPI

pip install submagic-mcp-server

From Source

git clone https://github.com/sidart10/submagic-mcp-server.git
cd submagic-mcp-server
python3 -m venv venv
source venv/bin/activate
pip install -r requirements.txt

Configuration

Get API Key

Sign up at https://app.submagic.co/signup to get your API key.

Claude Desktop

Add to your claude_desktop_config.json:

{
  "mcpServers": {
    "submagic": {
      "command": "python",
      "args": ["-m", "submagic_mcp"],
      "env": {
        "SUBMAGIC_API_KEY": "sk-your-api-key-here"
      }
    }
  }
}

If installed from source, use absolute paths:

{
  "mcpServers": {
    "submagic": {
      "command": "/path/to/venv/bin/python",
      "args": ["/path/to/submagic_mcp.py"],
      "env": {
        "SUBMAGIC_API_KEY": "sk-your-api-key-here"
      }
    }
  }
}

Restart Claude Desktop after configuration.

Tools

submagic_list_languages

Get list of supported languages for transcription.

Returns 107+ language codes (e.g., "en", "es", "fr", "cmn_en").

Rate limit: 1000 requests/hour

submagic_list_templates

Get available video styling templates.

Returns 30+ template names including Hormozi series, Beast, Sara, and others.

Rate limit: 1000 requests/hour

submagic_create_project

Create a new video project with AI features.

Inputs:

• title (string): Project title
• language (string): Language code from submagic_list_languages
• video_url (string): Public URL to video file
• template_name (string, optional): Template name
• user_theme_id (string, optional): Custom theme UUID
• magic_zooms (boolean, optional): Enable AI zooms (default: true)
• magic_brolls (boolean, optional): Enable auto B-rolls (default: true)
• magic_brolls_percentage (integer, optional): B-roll coverage 0-100 (default: 75)
• remove_silence_pace (string, optional): "natural", "fast", or "extra-fast" (default: "medium")
• remove_bad_takes (boolean, optional): Remove filler words (default: true)
• dictionary (array, optional): Custom words for transcription accuracy
• webhook_url (string, optional): Completion notification URL

Returns project ID and status. Processing takes 2-10 minutes.

Rate limit: 500 requests/hour

submagic_get_project

Get project details and status.

Inputs:

• project_id (string): UUID from submagic_create_project

Returns complete project information including status, settings, and download URL when complete.

Rate limit: 500 requests/hour

submagic_update_project

Update project settings after creation.

Inputs:

• project_id (string): Project UUID
• remove_silence_pace (string, optional): Adjust silence removal speed
• remove_bad_takes (boolean, optional): Enable/disable filler word removal
• custom_broll_items (array, optional): Insert custom B-roll clips
  • Each item: {startTime: float, endTime: float, userMediaId: string}

Must re-export project after updating to apply changes.

Rate limit: 100 requests/hour

submagic_export_project

Export completed project video.

Inputs:

• project_id (string): Project UUID
• fps (integer, optional): Frames per second 1-60 (default: 30)
• width (integer, optional): Video width in pixels 100-4000 (default: 1080)
• height (integer, optional): Video height in pixels 100-4000 (default: 1920)
• webhook_url (string, optional): Export completion notification URL

Returns export status. Use submagic_get_project to get download URL.

Rate limit: 500 requests/hour

submagic_create_magic_clips

Generate viral short-form clips from YouTube videos.

Inputs:

• title (string): Project title
• youtube_url (string): YouTube video URL
• language (string): Language code
• min_clip_length (integer, optional): Minimum duration in seconds 15-300 (default: 15)
• max_clip_length (integer, optional): Maximum duration in seconds 15-300 (default: 60)
• user_theme_id (string, optional): Custom theme UUID
• webhook_url (string, optional): Completion notification URL

AI analyzes video and generates multiple optimized clips. Processing takes 5-15 minutes.

Rate limit: 500 requests/hour

Usage Examples

Create Video with AI Captions

submagic_create_project(
    title="Product Demo",
    language="en",
    video_url="https://example.com/video.mp4",
    template_name="Hormozi 2",
    remove_silence_pace="fast"
)

Generate TikTok Clips from YouTube

submagic_create_magic_clips(
    title="TikTok Highlights",
    youtube_url="https://youtube.com/watch?v=abc123",
    language="en",
    min_clip_length=15,
    max_clip_length=30
)

Export for Instagram Reels (9:16)

submagic_export_project(
    project_id="project-uuid",
    width=1080,
    height=1920,
    fps=30
)

Limitations

• Videos must be publicly accessible URLs
• Maximum file size: 2GB
• Maximum duration: 2 hours
• Supported formats: MP4, MOV
• remove_bad_takes adds 1-2 minutes processing time
• Some dashboard features not available via API (see docs/API_LIMITATIONS_DISCOVERED.md)

Development

Run Tests

python tests/test_server.py

Project Structure

submagic-mcp-server/
├── submagic_mcp.py      # Main server implementation
├── tests/               # Test suite
├── docs/                # API documentation
├── pyproject.toml       # Package configuration
└── requirements.txt     # Dependencies

License

MIT

Resources

• API Documentation: https://docs.submagic.co
• Get API Key: https://app.submagic.co/signup
• MCP Protocol: https://modelcontextprotocol.io