No-Code Automation Using Playwright MCP Server
Let GitHub Copilot control your browser — no test code needed
"You write instructions in plain English. Playwright MCP reads them. A real browser opens and does exactly what you said. No code. No selectors. No setup headache."
I was pretty surprised when I first saw this work. You describe what you want a browser to do — in normal sentences — and GitHub Copilot + Playwright MCP Server just does it. A real Chrome window opens, goes to the right page, types in the right field. All from a plain text file.
This blog walks you through the whole thing from scratch. If you’ve never used any of these tools before, that’s fine. We start from zero.
What is the Playwright MCP Server?
Let’s break down the two parts first.
Playwright is a tool by Microsoft that controls web browsers using code. It can open Chrome, type things, click buttons, and check what’s on screen — all automatically.
MCP (Model Context Protocol) is a standard way for AI tools to talk to external tools and services. Think of it as a plug that connects your AI assistant to things outside it — like a browser.
Playwright MCP Server combines both. It’s a bridge that lets an AI assistant like GitHub Copilot actually drive a real browser through Playwright. You describe what to do in plain English. The AI reads it, talks to the MCP server, and Playwright carries it out in a real browser.
In short: Instead of you writing code to automate a browser, you write plain English instructions, and the AI + Playwright does the rest.
What tools do you need?
Before we set anything up, here’s everything you’ll need installed on your computer:
Tool | What it’s for | Where to get it |
Node.js (v18+) | Runs JavaScript tools on your computer | nodejs.org → download LTS |
VS Code | The code editor where everything happens | |
GitHub Copilot extension | The AI assistant inside VS Code | VS Code Extensions panel |
GitHub Copilot Chat extension | Lets you chat with Copilot and use Agent Mode | VS Code Extensions panel |
Playwright MCP Server | The bridge between Copilot and your browser | Installed in Step 3 below |
Step-by-step setup
Follow these steps in order. Each one builds on the previous.
Step 1 — Install Node.js
Go to nodejs.org and download the LTS version. LTS stands for “Long Term Support” — it’s the most stable one. Install it like any normal app.
Once it’s installed, open a terminal (on Mac: Terminal app, on Windows: Command Prompt or PowerShell) and type:
// Terminal
node --version
# You should see something like: v20.11.0
If you see a version number, Node.js is working.
Step 2 — Install VS Code and GitHub Copilot
• Download and install VS Code from code.visualstudio.com
• Open VS Code, click the Extensions icon on the left sidebar (it looks like 4 squares)
• Search for GitHub Copilot and install it
• Search for GitHub Copilot Chat and install it too
• Sign in with your GitHub account when prompted — you need a GitHub Copilot subscription for this to work
Step 3 — Install the Playwright MCP Server in VS Code
This is the key step. There are two ways to do it:
Option A — One-click install from GitHub (easiest):
1. Go to this URL in your browser: github.com/microsoft/playwright-mcp
2. Look for the button that says "Install server for VS Code" and click it
3. VS Code will open and ask you to confirm — click Allow
Option B — Manual config (if Option A doesn’t work):
4. In VS Code, press Ctrl+Shift+P (Windows) or Cmd+Shift+P (Mac) to open the Command Palette
5. Type >mcp and select MCP: Open User Configuration
6. A file called mcp.json will open. Paste this inside it:
mcp.json
{
"servers": {
"playwright": {
"command": "npx",
"args": ["@playwright/mcp@latest"],
"type": "stdio"
}
}}
}
7. Save the file
Step 4 — Verify the MCP server is connected
• Open Copilot Chat in VS Code — click the chat icon in the sidebar or press Ctrl+Alt+I
• In the chat panel, look for the mode selector at the top and switch it to Agent mode
• Click Configure Tools (or the tools icon) and check that playwright appears in the list
you see “playwright” in the tools list, you’re all set. The browser is now connected to your AI assistant.
Create your instructions file
Now here’s the fun part. You don’t write test code. You write a plain text file with instructions in normal English (or Markdown), and GitHub Copilot reads it.
Step 5 — Create the instructions.md file
8. In VS Code, create a new folder called my-automation
9. Inside it, create a new file called instructions.md
10. Paste the following into the file:
11. Save the file
That’s literally the whole “test script.” Two lines of plain English. No code. No selectors. No browser setup.
Run it using GitHub Copilot Agent Mode
Now you’ll use GitHub Copilot to read your instructions file and run the automation in a real browser.
12. Open Copilot Chat and make sure you’re in Agent mode (check the dropdown at the top of the chat panel)
13. Attach your instructions.md file to the chat. You can drag and drop it in, or click the paperclip/attachment icon
14. Type this message in the chat:
Copilot Chat — Agent Mode
Follow the steps in instructions.md and run them in a browser using Playwright MCP. 15. Press Enter and watch what happens:
• A real Chrome browser window opens automatically
• It navigates to google.com
• It types "numpyninja" into the search field
• Copilot reports back what it did in the chat window
Tip: If the browser doesn’t open, make sure you’re in Agent mode (not Ask or Edit mode) and that the playwright MCP tool is checked in the tools list.
What just happened? (The simple explanation)
If you’re wondering what actually happened behind the scenes:
• You wrote instructions in plain English in instructions.md
• GitHub Copilot (in Agent mode) read your file and understood what to do
• Copilot sent those instructions to the Playwright MCP Server
• The MCP Server told Playwright: open Chrome, go to this URL, type this text
• Playwright carried out the actions in a real browser
• Results came back to Copilot, which reported back to you in the chat
No test framework. No selectors. No JavaScript file. Just plain English — and a real browser that follows it.
Why is this useful?
This approach is great for a few different situations:
• Beginners who want to try automation without learning a programming language first
• Quick checks where you just need to verify something works and don’t want to write a full test
• Exploratory testing where the steps might change and you don’t want to maintain code
• Demos where you want to show someone how a web flow works
And once you’re comfortable with this, you can ask Copilot to turn your instructions.md into a proper Playwright test file — so you can graduate from no-code to real automation when you’re ready.
The bottom line
The Playwright MCP Server removes the biggest barrier for beginners: you don’t need to know how to code to automate a browser. You just need to know what you want the browser to do — and write it down like you’d explain it to a person.
Install Node.js, set up VS Code with Copilot, connect the MCP server, write your instructions.md, and hit send. That’s it.
