Building a YouTube MCP Server with MuleSoft and CurieTech AI in 10 Minutes

Curious how to integrate the YouTube Data API v3 into a fully functioning MCP (Model Context Protocol) server in record time? With the help of CurieTech AI and MuleSoft, I did it in just 10 minutes, without ever having created an MCP server before.

In this article, I’ll walk you through what I did step-by-step — including how I used CurieTech to generate the entire integration, converted it into an MCP server, and ran it using SuperGateway and the MCP Inspector. I also show how you can connect this to Claude or other apps and use natural language to run YouTube searches via the MCP server.

🔗 GitHub Repo: https://github.com/alexandramartinez/youtube-mcp-server-mule

🧠 What Is MCP?

MCP stands for Model Context Protocol, a new protocol designed for agents, AI tools, and other automated systems to interact with structured tools in a flexible and self-describing way. MCP tools define schemas, input parameters, and responses — and are ideal for agent-based use cases.

In other words… It’s an API for your agents.

🛠️ What We’re Building

A MuleSoft app that:

• Connects to the YouTube Data API v3
• Accepts search queries and API keys as parameters
• Returns YouTube search results
• Runs as an MCP server with full schema and parameter validation

🔑 Step 1: Get Your YouTube API Key

To use the YouTube API, you’ll need an API key from Google Cloud:

• Go to Google Cloud Console.
• Create a project.
• Enable the YouTube Data API v3.
• Under APIs & Services > Credentials, create a new API Key.
• Under API Restrictions, select YouTube Data API v3.
• Save your changes and copy the key — you’ll use it in your requests.

🧪 Step 2: Generate the API with CurieTech AI

Inside CurieTech AI:

• Go to the Integration Generator.
• Set the language to Java 17, build tool to Maven 3.9, and Mule 4.9.
• Use a prompt like:

“Make an API that uses YouTube Data API v3. Include all query parameters, especially key, and make it as simple as possible.”

• Paste the list of parameters from the official docs or paste the URL.
• CurieTech generates the full integration — copy the code.

💻 Step 3: Run the App in MuleSoft

• Create a new Mule project in Anypoint Code Builder or Studio.
• Paste the code from CurieTech.
• Run the project (default port is 8081).
• Use Postman to test:

GET http://localhost:8081/search?key=YOUR_API_KEY&q=mulesoft

🤖 Step 4: Convert to an MCP Server

• Push your Mule project to GitHub (or use VS Code’s repo initializer).
• In CurieTech, open the Coding Agent and select your repo.
• Prompt:

“Turn this into an MCP server using the beta MCP connector.”

• CurieTech updates the code, modifies the pom.xml, adds a global MCP config, and replaces the HTTP listener with an MCP listener.
• Approve the PR in GitHub and merge.

🚦 Step 5: Run SuperGateway

• Open a terminal window.
• Run the following command (ensure the port matches your Mule app’s HTTP Listener):

npx -y supergateway --sse http://localhost:8081/sse --ssePath /sse --messagePath /message

🔍 Step 6: Open MCP Inspector

• Open a new terminal window and run:

npx @modelcontextprotocol/inspector

• Copy the tokenized URL (e.g., http://localhost:6274/?MCP_PROXY_AUTH_TOKEN=…) into your browser.
• Select SSE as the transport type
• Paste the following URL:

http://0.0.0.0:8081/sse

• Click Connect
• Select the youtube-search tool
• Enter your API Key and query
• Click on Run Tool

🤝 Step 7: Add the Server to Your App (e.g., Claude)

Add the following to your Claude or other AI agent config:

{
  "mcpServers": {
    "youtube": {
      "command": "npx",
      "args": [
        "-y",
        "supergateway",
        "--sse",
        "http://0.0.0.0:8081/sse",
        "--ssePath",
        "/sse",
        "--messagePath",
        "/message"
      ]
    }
  }
}

Restart your app, and you’re ready to use it in natural language prompts!

🧠 Example Prompts You Can Use

Inside Claude (or any compatible tool):

• “Search for the ProstDev YouTube channel and return the last 4 videos.”
• “Generate a LinkedIn post that promotes the first video. Include the title, description, and video link.”
• “Find trending content about ‘API design’ using YouTube search.”

📦 Repo

🔗 GitHub Repo: https://github.com/alexandramartinez/youtube-mcp-server-mule

🧭 Final Thoughts

I had never worked with MCP before — and I built this whole server in 10 minutes thanks to CurieTech AI. The entire flow, from integration to deployment and testing, was smoother than I imagined. If you learn by reading code like I do, this is one of the fastest ways to master new patterns like MCP.

Whether you’re building for fun, research, or AI-driven platforms, give this a try and let me know what you build next!