Based on a tutorial by Trại Ey Eye
If you’ve been curious about MCP (Model Context Protocol) servers but felt intimidated by the coding requirements, you’re not alone. Many tutorials assume programming knowledge, leaving non-coders feeling left out of this powerful technology.
I’m summarizing this excellent Vietnamese tutorial that shows exactly how to create a fully functional MCP server without writing a single line of code. The creator demonstrates using AI agents to do all the heavy lifting while you focus on defining what you want your server to accomplish.
Quick Navigation
- What You’ll Learn (00:00-02:30)
- Setting Up Your Environment (02:31-05:00)
- Configuring the AI Agent (05:01-08:30)
- Creating the MCP Server (08:31-12:00)
- Testing Your Server (12:01-15:30)
- Validating with n8n (15:31-18:00)
What You’ll Learn
This tutorial addresses a gap in the MCP learning landscape. While many videos explain MCP concepts and show coding examples, they’re typically aimed at developers.
Key Points:
- Create a functional MCP server without any programming knowledge
- Use AI agents for “live coding” – having AI write code in real-time
- Build a server that reads Excel/CSV files and returns data analysis
- Test and validate your server using automation tools
My Take:
The democratization of server creation through AI is remarkable. This approach removes technical barriers while still teaching fundamental concepts about how MCP servers function.
Setting Up Your Environment
Before diving into server creation, you’ll need the right tools and reference materials. The tutorial emphasizes preparation as crucial for success.
Required Downloads:
- IDE with AI Support: VS Code, Cursor, or Windsurf (tutorial uses VS Code)
- MCP Server Examples: Download the official example files from Anthropic
- Programming Guide: Get the README file for either TypeScript or Python (tutorial chooses Python)
The creator specifically recommends downloading these reference materials because they’ll serve as the knowledge base for your AI agent. Think of them as giving your AI assistant a comprehensive manual to work from.
My Take:
This preparation step is often overlooked but absolutely critical. The quality of your AI agent’s output directly correlates with the quality of reference materials you provide.
Configuring the AI Agent
With your environment ready, it’s time to configure your AI assistant. This involves selecting the right model and providing proper context.
Configuration Steps:
- Ensure VS Code is set to use an AI assistant (the tutorial uses Claude 3.5)
- Add both downloaded files to the AI’s context using the “Add Context” button
- Verify the AI has access to MCP examples and programming guidelines
The tutorial emphasizes choosing a capable model like Claude 3.5 Sonnet for best results. The AI needs enough context to understand both MCP concepts and Python implementation details.
My Take:
Context management is crucial here. By loading both the examples and documentation, you’re essentially giving the AI a complete reference library to draw from during code generation.
Creating the MCP Server
This is where the magic happens. Using a carefully crafted prompt, you’ll direct the AI to build your entire MCP server.
The Prompt Strategy:
- Part 1: Define what your server should do (read Excel/CSV files, return row counts and column names)
- Part 2: Provide step-by-step instructions for creating and programming the MCP server
- Execution: Simply paste the prompt and let the AI work
Example server functionality:
- Access local data folder
- Read Excel or CSV files
- Return number of rows
- List all column names
The beauty of this approach is its simplicity. You describe what you want in plain language, and the AI translates that into functional Python code that implements the MCP protocol.
My Take:
The prompt engineering here is elegant in its simplicity. By clearly separating the “what” (functionality) from the “how” (implementation steps), you guide the AI to produce clean, purposeful code.
Testing Your Server
Once the AI generates your server code, testing is straightforward. You don’t need to understand the code to verify it works.
Running the Server:
- Simply type “run the server” in your AI chat
- The AI will execute the necessary commands
- Verify the server starts without errors
- Note the server address (typically localhost:8000)
The tutorial shows the server starting successfully, confirming that the AI-generated code is functional. This immediate feedback loop helps build confidence in the process.
My Take:
This testing approach removes another barrier for non-programmers. You don’t need to understand terminal commands or debugging – just ask the AI to run what it created.
Validating with n8n
To prove the server works end-to-end, the tutorial demonstrates connecting it to n8n, a popular automation platform.
n8n Integration:
- Create a simple AI agent in n8n
- Connect it to Google Gemini for AI capabilities
- Add MCP client connection pointing to your server (localhost:8000/sse)
- Test with a simple query like “analyze test.csv”
The test demonstrates real functionality: the system correctly identifies 148 rows and lists all column names (PassengerID, PassName, etc.) from the test CSV file. This proves the entire pipeline works.
My Take:
Using n8n for validation is brilliant because it simulates real-world usage. You’re not just running isolated tests – you’re proving your server can integrate with actual automation workflows.
Key Takeaways
This tutorial proves that MCP server creation doesn’t require programming expertise. By leveraging AI agents and proper preparation, anyone can build functional servers for their specific needs.
Success Factors:
- Preparation: Download proper reference materials
- Context: Provide comprehensive documentation to your AI
- Clear Prompts: Separate functionality from implementation
- Testing: Validate with real-world tools
The approach demonstrates how AI can democratize technology creation, making powerful tools accessible to non-technical users while maintaining professional-grade results.
