BrowserStack MCP

Connect AI agents like GitHub Copilot, Cursor, or Claude to the full power of BrowserStack’s Test Platform. This Model Context Protocol (MCP) server lets you test, debug, and manage web and mobile apps across 3,000+ real browsers and devices—just by using natural language prompts.

What is the BrowserStack MCP Server?

The BrowserStack MCP Server is a Model Context Protocol (MCP) server that acts as a bridge between BrowserStack’s cloud-based testing infrastructure and AI-enabled clients like GitHub Copilot, Cursor, and Claude.

What Can You Do with It?

• Manual Testing

  Instantly open and test your website or mobile app on real devices (like iPhone 15 Pro or Android 14) without needing emulators or local setup.

• Automated Testing

  Run and debug Selenium, Playwright, or Jest tests directly from your IDE. Fix flaky tests, port your suite to the cloud, or investigate CI failures—all through AI-driven prompts.

• Accessibility Scanning

  Scan local or production websites for accessibility issues (WCAG, ADA compliance) and get AI-suggested fixes with no extra tooling.

• Test Management

  Create projects, add test cases, trigger test runs, and update results—all via natural language. Fully integrated with BrowserStack’s Test Management features.

• AI Agent Integrations

  Generate test cases from PRDs, convert manual to automated tests, and auto-heal scripts using BrowserStack’s own AI agents.

All this is done directly from your IDE or AI assistant, with no need to switch contexts or write complex scripts.

When Should You Use This?

Use the BrowserStack MCP Server if:

• You want to control BrowserStack from inside your IDE or AI assistant
• You’re building or debugging test suites and want to run them in the cloud
• You need to test across multiple browsers, OS versions, or real devices
• You’re tired of switching between tools for test case management and execution
• You want to leverage AI for writing, debugging, and healing tests

Installation Guide

Prerequisites:

• Node.js ≥ 18.0 (Recommended: v22.15.0 LTS)
• BrowserStack account with Access Key

Steps:

1. Create a BrowserStack account

   Sign up here. Open-source projects may qualify for free access.

2. Configure MCP Clients

   • GitHub Copilot / VS Code

       json
       CopyEdit
       {
         "servers": {
           "browserstack": {
             "command": "npx",
             "args": ["-y", "@browserstack/mcp-server@latest"],
             "env": {
               "BROWSERSTACK_USERNAME": "<username>",
               "BROWSERSTACK_ACCESS_KEY": "<access_key>"
             }
           }
         }
       }
     

   • Cursor IDE

       json
       CopyEdit
       {
         "mcpServers": {
           "browserstack": {
             "command": "npx",
             "args": ["-y", "@browserstack/mcp-server@latest"],
             "env": {
               "BROWSERSTACK_USERNAME": "<username>",
               "BROWSERSTACK_ACCESS_KEY": "<access_key>"
             }
           }
         }
       }
     

   • Claude Desktop (via Smithery)

       bash
       CopyEdit
       npx -y @smithery/cli install @browserstack/mcp-server --client claude
     

3. Start the MCP Server

   In your IDE, click “Start” to launch the BrowserStack server and begin issuing commands through your AI assistant.

Recommended Clients

• GitHub Copilot and Cursor: Ideal for AI-powered automated testing and debugging
• Claude Desktop: Best for manual testing and exploratory workflows