home / mcp / chrome mcp server

Chrome MCP Server

Provides a local MCP bridge between Chrome DevTools and Cursor AI for logs, network data, screenshots, and element inspection.

Installation
Add the following to your MCP client configuration file.

Configuration

View docs
{
  "mcpServers": {
    "adejaimejr-chrome-mcp-server": {
      "command": "npx",
      "args": [
        "github:adejaimejr/chrome-mcp-server"
      ]
    }
  }
}

The Chrome MCP Server provides a local bridge between the Chrome DevTools extension and Cursor AI. It captures console logs and errors, monitors network activity, takes screenshots, inspects elements, and clears logs, so you can debugging and monitor web pages directly from Cursor.

How to use

You use the MCP server by starting it on your machine and connecting the Cursor MCP client to it. Choose a command method that fits your workflow, then configure Cursor to point at the running server. The server exposes tools for collecting logs, network data, screenshots, and element information, so you can inspect and analyze browser behavior from Cursor.

How to install

Prerequisites: you need Node.js 14+ and npm installed on your system.

Option 1 — Run directly from the source with npx:

Command to start directly from the source: npx github:adejaimejr/chrome-mcp-server

Option 2 — Install globally and run the command directly after installation:

npm install -g github:adejaimejr/chrome-mcp-server

chrome-mcp-server

Option 3 — Publish on npm and run via npx (choose unscoped or scoped package as shown):

# Unscoped package
npm login
npm publish

# Then you can run:
npx chrome-mcp-server

# Scoped package
# Modify package.json name to "@adejaimejr/chrome-mcp-server" and publish publicly
npm publish --access public

# Then you can run:
npx @adejaimejr/chrome-mcp-server

Option 4 — Run a local copy if you already have the code in a local path (Windows example shown): You can execute the built file directly with Node.

node C:\Users\Adejaime\Desktop\2025-dev\chrome-mcp-server\dist\mcp-server.js

Option 5 — Clone, build, and run locally from source (recommended for development):

git clone https://github.com/adejaimejr/chrome-mcp-server.git
cd chrome-mcp-server
npm install
npm run build

# Configure Cursor with the absolute path to the built file, e.g. on Windows:
# node C:\path\to\chrome-mcp-server\dist\mcp-server.js

# On Linux/macOS:
# node /path/to/chrome-mcp-server/dist/mcp-server.js

Additional configuration and notes

Cursor can be configured to connect to one or more MCP servers. Open Cursor, go to Settings, then Extensions > MCP, and add a new MCP server with one of the following runtime commands.

If you install via npx, you can use the following start commands in Cursor configuration:

# Unscoped package
npx chrome-mcp-server

# Scoped package
npx @adejaimejr/chrome-mcp-server

If you installed globally, Cursor can point at the local global command directly by using the executable name:

chrome-mcp-server

If you run a local build, specify the absolute path to the built script in Cursor, for example on Windows or Linux/macOS as shown earlier.

Note: The most recent version automatically detects when it is running under Cursor and does not require a --mcp flag. It also attempts to bind to an available port if the default port is in use and reports the active port to Cursor.

Security and usage notes

This server is intended for local development use only. Do not expose it to public networks without proper security controls.

Troubleshooting and tips

If you encounter JSON parsing errors, ensure you are using a recent version that guarantees JSON-formatted output. If port conflicts occur, the server will try alternate ports automatically and report the port it is using.

What you can do with the MCP server

The server provides a set of functions to collect and manage debugging data from the browser through the MCP interface, including retrieving console logs, console errors, network logs, taking screenshots, inspecting the selected element, and clearing logs.

Available tools

getConsoleLogs

Fetches logs collected from the browser's console output.

getConsoleErrors

Fetches errors captured from the browser console.

getNetworkErrorLogs

Fetches network request errors captured during page loads.

getNetworkSuccessLogs

Fetches successful network request records.

takeScreenshot

Captures a screenshot of the current browser page.

getSelectedElement

Retrieves information about the element currently selected in the browser.

wipeLogs

Clears all stored logs and data.