Is it just us or is everyone talking about MCP (Model Context Protocol) these days?

If you’ve been smiling, nodding, and hoping no one asks your opinion—but really need the TL;DR on MCP—here’s the deal:

LLMs often struggle when they’re asked for information outside of their training data. Sometimes they’ll hallucinate and say something incorrect, or simply say, “I don’t know.” For AI agents built on top of LLMs to be useful, you’ll need to provide the right amount of context when you prompt them, whether it’s your codebase, your repository data, your documentation, or something else.

This is where MCP comes in. It’s an open and secure standard for connecting LLMs and AI assistants to real-world tools and data.

🤖 Want to see it in action? Here are three ways to use MCP today:

1. Use MCP to bridge design and development teams with Figma 🤝

MCP gives GitHub Copilot a standardized way to pull exact design details—like spacing, color, and component states—directly from Figma. This allows it to generate clean, ready-to-use code without manual translation. Here’s how to give it a try:

Open the Figma desktop app
Go to: Figma → Preferences → Enable Dev Mode MCP Server

Configure your editor (e.g., VS Code)
Open settings.json and add:

Select a frame in Figma
(e.g., a button group or pricing card)

Open Copilot Chat and prompt it with:

That’s it
Copilot pulls the design context and turns it into functional code, no guesswork required.

👉 Read the full Figma MCP server guide >

2. File pull requests faster ⚡

GitHub’s remote MCP server, now in public beta for VS Code and Visual Studio, gives tools like Copilot, Cursor, and Windsurf secure, real-time access to your GitHub data. This means you can file pull requests, retrieve files, and reference issues without ever leaving your editor. No setup required! Here’s how to get started:

Install the Remote MCP Server

• Go to the GitHub MCP Server repository
• One-click install into VS Code OR Copy the remote server URL and paste it into any MCP-compatible host

Authenticate

• Sign in with OAuth 2.0 (recommended) or use a PAT with the right scopes
• If you’re using VS Code, make sure your org has enabled the Editor Preview Policy

Prompt Copilot in chat:

That’s it
Copilot will query GitHub via MCP, create the file, commit it, and open the pull request—without the need for context switching.

👉 Read the full MCP server changelog >

3. Monitor application performance with Grafana 👀

Want to keep tabs on your systems with just a prompt? The open source Grafana MCP server connects GitHub Copilot to Grafana so you can monitor logs, metrics, and alerts in real time, no dashboard digging required. Here’s a simple way to try it:

Run the Grafana MCP server (Docker SSE mode)

Configure VS Code
Add the following to .vscode/settings.json:

Prompt Copilot to check performance

That’s it
Copilot pulls live metrics via MCP, so you can identify issues, check alerts, or query logs on the fly.

👉 Read the full Grafana MCP server guide >

Now the next time someone brings up MCP in a conversation, you don’t have to fake it. You’ll even have your own use case to show off.

Happy coding! 👩‍💻👨‍💻

---

✨ This newsletter was produced by Gwen Davis. ✨

---

The world’s leading AI-powered developer platform.