Video-Tool-MCP

johnsonboibeats/Video-Tool-MCP

3.1

If you are the rightful owner of Video-Tool-MCP and would like to certify it and/or have it hosted online, please leave a comment on the right or send an email to dayong@mcphub.com.

A comprehensive Model Context Protocol (MCP) server for video generation and processing tasks, deployed on Railway with Claude Desktop compatibility and full OAuth 2.0 support.

Tools
5
Resources
0
Prompts
0

Video Tool MCP Server

A comprehensive Model Context Protocol (MCP) server for video generation and processing tasks, deployed on Railway with Claude Desktop compatibility and full OAuth 2.0 support.

Features

  • Video Generation: Create videos from text prompts using Google Veo3 API
  • Video Analysis: Analyze videos with detailed descriptions using GPT-4o
  • Video Editing: Edit videos with text prompts and optional masks
  • Video Variations: Generate variations of existing videos
  • Text Extraction: Extract text from video frames using OCR
  • Video Comparison: Compare two videos for similarities and differences
  • Smart Editing: Advanced editing with analysis and modification prompts
  • Video Transformations: Resize, rotate, flip, and apply filters to video frames
  • Batch Processing: Process multiple videos with various operations
  • Video Metadata: Extract technical information from videos
  • Prompt Generation: Generate optimized prompts from video content
  • Local File Support: Process files from local file system
  • Base64 Support: Handle base64 encoded video data
  • Google Drive Integration: Search, upload, and process videos directly from Google Drive
  • OAuth 2.0 Support: Full OAuth implementation for Claude integration

OAuth 2.0 Implementation

This server includes a complete OAuth 2.0 implementation that is fully compliant with Anthropic's requirements for remote MCP servers.

OAuth Features

  • ✅ Authorization Code Grant flow
  • ✅ Refresh Token support
  • ✅ Token revocation
  • ✅ OAuth discovery endpoints
  • ✅ User info endpoint
  • ✅ Claude-compatible error responses
  • ✅ JWT token handling
  • ✅ Secure token storage

OAuth Endpoints

  • /.well-known/oauth-authorization-server - OAuth server discovery
  • /.well-known/oauth-protected-resource/mcp - Protected resource info
  • /oauth/authorize - Authorization endpoint
  • /oauth/token - Token exchange endpoint
  • /oauth/userinfo - User information endpoint
  • /oauth/revoke - Token revocation endpoint
  • /oauth/test - OAuth testing endpoint
  • /oauth/status - OAuth server status

OAuth Configuration

  • Supported Grant Types: Authorization Code, Refresh Token
  • Supported Scopes: read, write, image_processing
  • Token Expiry: Access tokens (1 hour), Refresh tokens (30 days)
  • Algorithm: HS256 (HMAC with SHA-256)

For detailed OAuth documentation, see .

Environment Variables

Required

  • GEMINI_API_KEY: Your Gemini API key for Veo3 video generation
  • OPENAI_API_KEY: Your OpenAI API key for video analysis (used by most tools)
  • GOOGLE_OAUTH_TOKEN: OAuth token JSON for Google Drive access (videos are auto-uploaded to Drive)

Optional

Model Configuration (Standardized Format)

The server supports a standardized provider:model-name format for specifying models:

  • CREATE_VIDEO_MODEL: Default model for video generation tools. Examples:
    • gemini:veo-3.0-generate-preview (Gemini API - default)
    • gemini:veo-3.0-fast-generate-preview (Gemini API)
    • gemini:veo-2.0-generate-001 (Gemini API)
    • Legacy format still supported: veo-3.0-generate-preview, veo-3.0-fast-generate-preview, etc.

Model Format: Use provider:model-name format for explicit provider selection, or legacy model names for backwards compatibility.

Note: Video generation uses Gemini API. Only Gemini provider is supported for video models. Other tools continue using OpenAI for analysis.

Server Configuration (Optional)
  • PORT: Server port (default: 8080)
  • ALLOWED_ORIGINS: CORS allowed origins (default: claude.ai domains)
  • MAX_REQUESTS_PER_MINUTE: Rate limiting (default: 100)
  • JWT_SECRET: Secret key for JWT signing (auto-generated if not provided)
  • JWT_EXPIRY_HOURS: Access token expiry in hours (default: 1)
  • REFRESH_TOKEN_EXPIRY_DAYS: Refresh token expiry in days (default: 30)

Local Development

Prerequisites

  • Python 3.8+
  • Google API key with Veo3 access
  • OpenAI API key for video analysis

Setup

  1. Clone the repository
  2. Install dependencies: pip install -r requirements.txt
  3. Set environment variables: 
    export GEMINI_API_KEY="your-gemini-api-key"
export OPENAI_API_KEY="your-openai-api-key"
export GOOGLE_OAUTH_TOKEN="your-oauth-token-json"
  4. Run the server: python server.py

Testing OAuth

Run the OAuth test script to verify the implementation:

python test_oauth.py

Railway Deployment

Environment Variables

{
  "GEMINI_API_KEY": "your-gemini-api-key",
  "OPENAI_API_KEY": "your-openai-api-key",
  "GOOGLE_OAUTH_TOKEN": "your-oauth-token-json",
  "JWT_SECRET": "your-jwt-secret-key"
}

Deploy

railway up

Claude Integration

Adding to Claude

  1. In Claude Desktop, go to Settings > Connectors
  2. Add your server URL (e.g., https://your-server.railway.app)
  3. Claude will automatically discover OAuth endpoints
  4. Complete the OAuth authorization flow
  5. Start using video processing tools

OAuth Flow

  1. Claude requests authorization via /oauth/authorize
  2. Server generates authorization code
  3. Claude exchanges code for access token
  4. Claude uses Bearer token for API requests
  5. Tokens are automatically refreshed when needed

Usage Examples

Video Generation

{
  "prompt": "A serene mountain landscape at sunset with flowing water",
  "duration": 5,
  "resolution": "1080p",
  "output_format": "mp4"
}

Video Analysis

{
  "video": "/path/to/video.mp4",
  "prompt": "Describe this video in detail"
}

Video Editing

{
  "video": "/path/to/video.mp4",
  "prompt": "Add a red car to the scene",
  "output_format": "mp4"
}

File Input Types

The server supports the following input types:

Local Files
{
  "video": "/absolute/path/to/video.mp4"
}
Base64 Data
{
  "video": "data:video/mp4;base64,AAAAIGZ0eXBpc29tAAACAGlzb21pc28y..."
}

API Endpoints

Health and Status

  • GET /health - Health check with OAuth status
  • GET / - Root endpoint with server info
  • GET /oauth/status - OAuth server status

OAuth Endpoints

  • GET /.well-known/oauth-authorization-server - OAuth discovery
  • GET /.well-known/oauth-protected-resource/mcp - Protected resource info
  • GET /oauth/authorize - Authorization endpoint
  • POST /oauth/token - Token exchange
  • GET /oauth/userinfo - User information
  • POST /oauth/revoke - Token revocation
  • GET /oauth/test - OAuth testing
  • GET /.well-known/jwks.json - JWKS endpoint

MCP Protocol

  • POST /mcp/ - MCP protocol endpoint

Available Tools

Video Processing

  1. create_video - Generate videos from text prompts using Google Veo3
  2. analyze_video - Analyze videos with detailed descriptions
  3. edit_video - Edit videos with text prompts
  4. generate_variations - Create variations of existing videos
  5. extract_text - Extract text from video frames using OCR
  6. batch_process - Process multiple videos
  7. video_metadata - Extract video metadata

Google Drive Integration

  1. search_videos - Search for video files in Google Drive
  2. upload_video - Upload videos to Google Drive with metadata
  3. get_video_from_drive - Get direct download URLs for Drive videos

Error Handling

The server includes comprehensive error handling for:

  • Invalid file paths
  • Missing API keys
  • Rate limiting
  • Network timeouts
  • Invalid video formats
  • OAuth authentication errors
  • Token validation failures

Security Features

  • Path traversal protection
  • Rate limiting
  • CORS configuration
  • Input validation
  • Secure file handling
  • OAuth 2.0 authentication
  • JWT token security
  • Token revocation support

Testing

OAuth Testing

python test_oauth.py

Manual Testing

# Test OAuth discovery
curl https://your-server.railway.app/.well-known/oauth-authorization-server

# Test health endpoint
curl https://your-server.railway.app/health

# Test OAuth status
curl https://your-server.railway.app/oauth/status

Production Considerations

OAuth Security

  • Use HTTPS in production
  • Set a strong JWT_SECRET
  • Consider using Redis for token storage
  • Implement rate limiting on OAuth endpoints
  • Monitor OAuth usage patterns

Token Storage

For production deployments, consider replacing in-memory token storage with:

  • Redis for distributed deployments
  • Database for persistent storage
  • Secure key management for JWT secrets

License

MIT License