Content Creator MCP Server

An MCP server designed for content creation tools. It provides tools to extract the spoken script/transcript from YouTube/Instagram videos using yt-dlp and Whisper, robust web article scraping with browser spoofing using curl_cffi + Jina Reader, and DuckDuckGo web searching using ddgs.

Prerequisites

1. Python 3.10+

2. FFmpeg installed and available in your system's PATH.

Related MCP server: YouTube Transcript Fetcher (YTT)

Installation

1. Create a virtual environment:

   python -m venv .venv

2. Activate the virtual environment:

   # On Windows PowerShell
   .\.venv\Scripts\Activate.ps1

3. Install dependencies:

   pip install -r requirements.txt

Running the Server

To test the server locally with the MCP Inspector:

mcp dev server.py

Integration: Google ADK (Agent Development Kit)

You can connect this MCP server to a Google ADK agent natively in Python:

import os
from google_genai.types import StdioServerParameters, StdioConnectionParams, McpToolset # Update your specific ADK imports

mcp_server_dir = r"C:\Users\Asus\Documents\V1_DOING_INTERSETING_INTERSETING_THINGS\ai\mcp-server-for-content-creatation"
python_exe = os.path.join(mcp_server_dir, ".venv", "Scripts", "python.exe")
server_script = os.path.join(mcp_server_dir, "server.py")

# Create the MCP Toolset
content_mcp_toolset = McpToolset(
    connection_params=StdioConnectionParams(
        server_params=StdioServerParameters(
            command=python_exe,
            args=[server_script],
            env={"PYTHONUNBUFFERED": "1", **os.environ}
        ),
        timeout=120.0  # Increased timeout for loading heavy ML libraries like Whisper
    )
)

Integration: Claude Desktop

Update your claude_desktop_config.json:

{
  "mcpServers": {
    "content-creator": {
      "command": "C:\\path\\to\\project\\.venv\\Scripts\\python.exe",
      "args": ["-m", "mcp", "run", "C:\\path\\to\\project\\server.py"]
    }
  }
}

Features

• extract_video_script: Given a valid URL (YouTube, Instagram), it downloads only the optimal audio stream, converts it to mp3, and transcribes the speech to text using the Whisper base model.

• fetch_content: Extracts the main readable text content from a generic web page. Equipped with curl_cffi to spoof Chrome TLS fingerprints avoiding Cloudflare Recaptcha blocks, and automatically routes through r.jina.ai as a headless-browser proxy ultimate fallback.

• web_search: Queries DuckDuckGo for web search results straight from your agent.