Hello, everyone! Today, I'm excited to share a cool AI feature that allows you to control your computer with just text prompts, even without relying on fancy AI agent tools. Imagine asking Claude-sonnet-3.7 to take a screenshot of the government bond yield from the Ministry of Finance website, and it does it perfectly!

This is made possible by the Model Context Protocol (MCP), a tool that enables large language models to interact with external applications and services. MCP is like a USB port for AI models, allowing them to connect with various "devices" (services) to perform tasks beyond their capabilities.

So, what is MCP? It's an open protocol developed by Anthropic that standardizes how applications provide context to large language models. Think of it as a common language that enables different AI models to communicate with various services.

Why do we need MCP? Just like how USB ports standardized device connections, MCP provides a standardized way for AI models to interact with external services, making it easier to integrate different applications and services.

Now, let's dive into how to use MCP services. First, you need to choose a client that supports MCP, such as Cherry Studio, Claude Desktop, or Cline. Then, select an MCP service you want to use, like Playwright, and follow the configuration instructions.

In Cherry Studio, you can add an MCP server by filling in the required parameters. Once configured, you can use the MCP service in the conversation window. For example, you can ask Claude to navigate to a website, take a screenshot, and save it as an image.

That's it for today's tutorial on using MCP services! Try it out and explore the possibilities of this new world. If you're interested in learning more, check out our previous tutorials on Workflow compilation, Prompt design, AI tool guides, and model updates.