Using the built-in mechanism of MCP in Github Copilot, you can extend Large Language Models (LLMs) the ability to automatically

• fetch latest information, such has from websites,
• update data into systems, such as write, delete or update records into files, databases, or application systems
• plan, orchestrate and schedule tasks or actions

Follow the following steps in the AI Agents Developer Productivity workshop

1. Visit the Github MCP Registry
2. Install the following MCP Services: Github, Playwrite,

Exercise 1: Test Github MCP Server

Once you have installed the Github MCP Server from the Registry, type the following in the prompt

1Use the Github MCP Server to study this code repository: 
2https://github.com/gocallum/ragfood and give me a short overview of what this code does?

The agent actually enters the repository and can:

• Explore directory structure and file relationships
• Read source code, configs, and documentation
• Analyze dependencies, imports, and data flows
• Cross-reference between files to understand architecture
• Identify patterns, anti-patterns, and specific implementations

You can even ask AI to study the code, learn from the patterns in the code, and apply this to your specific problem area.

Exercise 2: Web Scraping with Playwright MCP Server

This tutorial shows the absolute basics of using the Playwright MCP server to interact with a live website and extract information.

1Use the Playwrite MCP server, Go to google.com, search for "weather in Sydney", and tell me what the weather today and next 3 days

You can just about go to any website, and let the agentic LLM to navigate the website, and perform tasks on your behalf.