YT to LinkedIn MCP Server

A Model Context Protocol (MCP) server that automates generating LinkedIn post drafts from YouTube videos. This server provides high-quality, editable content drafts based on YouTube video transcripts.

Features

• YouTube Transcript Extraction: Extract transcripts from YouTube videos using video URLs
• Transcript Summarization: Generate concise summaries of video content using OpenAI GPT
• LinkedIn Post Generation: Create professional LinkedIn post drafts with customizable tone and style
• Modular API Design: Clean FastAPI implementation with well-defined endpoints
• Containerized Deployment: Ready for deployment on Smithery

Setup Instructions

Prerequisites

• Python 3.8+
• Docker (for containerized deployment)
• OpenAI API Key
• YouTube Data API Key (optional, but recommended for better metadata)

Local Development

1. Clone the repository:

   git clone <repository-url>
   cd yt-to-linkedin

2. Create a virtual environment and install dependencies:

   python -m venv venv
   source venv/bin/activate  # On Windows: venvScriptsactivate
   pip install -r requirements.txt

3. Create a .env file in the project root with your API keys:

   OPENAI_API_KEY=your_openai_api_key
   YOUTUBE_API_KEY=your_youtube_api_key

4. Run the application:

   uvicorn app.main:app --reload

5. Access the API documentation at http://localhost:8000/docs

Docker Deployment

1. Build the Docker image:

   docker build -t yt-to-linkedin-mcp .

2. Run the container:

   docker run -p 8000:8000 --env-file .env yt-to-linkedin-mcp

Smithery Deployment

1. Ensure you have the Smithery CLI installed and configured.

2. Deploy to Smithery:

   smithery deploy

API Endpoints

1. Transcript Extraction

Endpoint: /api/v1/transcript
Method: POST
Description: Extract transcript from a YouTube video

Request Body:

{
  "youtube_url": "https://www.youtube.com/watch?v=VIDEO_ID",
  "language": "en",
  "youtube_api_key": "your_youtube_api_key"  // Optional, provide your own YouTube API key
}

Response:

{
  "video_id": "VIDEO_ID",
  "video_title": "Video Title",
  "transcript": "Full transcript text...",
  "language": "en",
  "duration_seconds": 600,
  "channel_name": "Channel Name",
  "error": null
}

2. Transcript Summarization

Endpoint: /api/v1/summarize
Method: POST
Description: Generate a summary from a video transcript

Request Body:

{
  "transcript": "Video transcript text...",
  "video_title": "Video Title",
  "tone": "professional",
  "audience": "general",
  "max_length": 250,
  "min_length": 150,
  "openai_api_key": "your_openai_api_key"  // Optional, provide your own OpenAI API key
}

Response:

{
  "summary": "Generated summary text...",
  "word_count": 200,
  "key_points": [
    "Key point 1",
    "Key point 2",
    "Key point 3"
  ]
}

3. LinkedIn Post Generation

Endpoint: /api/v1/generate-post
Method: POST
Description: Generate a LinkedIn post from a video summary

Request Body:

{
  "summary": "Video summary text...",
  "video_title": "Video Title",
  "video_url": "https://www.youtube.com/watch?v=VIDEO_ID",
  "speaker_name": "Speaker Name",
  "hashtags": ["ai", "machinelearning"],
  "tone": "professional",
  "voice": "first_person",
  "audience": "technical",
  "include_call_to_action": true,
  "max_length": 1200,
  "openai_api_key": "your_openai_api_key"  // Optional, provide your own OpenAI API key
}

Response:

{
  "post_content": "Generated LinkedIn post content...",
  "character_count": 800,
  "estimated_read_time": "About 1 minute",
  "hashtags_used": ["#ai", "#machinelearning"]
}

4. Output Formatting

Endpoint: /api/v1/output
Method: POST
Description: Format the LinkedIn post for output

Request Body:

{
  "post_content": "LinkedIn post content...",
  "format": "json"
}

Response:

{
  "content": {
    "post_content": "LinkedIn post content...",
    "character_count": 800
  },
  "format": "json"
}

Environment Variables

Note: While environment variables for API keys are optional (as they can be provided in each request), it's recommended to set them for local development and testing. When deploying to Smithery, users will need to provide their own API keys in the requests.

License

MIT