BrowserStack MCP Server

What is BrowserStack MCP Server?

The BrowserStack MCP Server connects the BrowserStack Test Platform to MCP-enabled clients like Cursor, VS Code, Claude Desktop, and Cline. It lets users manage test cases, execute manual or automated tests, debug issues, and fix code using natural language prompts, reducing context switching.

How to use BrowserStack MCP Server?

Install via one-click buttons for VS Code or Cursor. Alternatively, set up manually by adding a configuration file (e.g., .vscode/mcp.json) with npx -y @browserstack/mcp-server@latest as the command, providing your BrowserStack username and access_key as environment variables. Requires Node.js version >=18.0.

Key features of BrowserStack MCP Server

• 20 tools for test management, execution, and debugging.
• Run manual tests on real devices and browsers via natural language.
• Execute and debug automated tests (Playwright, Selenium, etc.).
• Scan accessibility issues on local or production sites.
• Access BrowserStack AI agents for test generation and auto-healing.
• Test localhost websites without remote deployment.

Use cases of BrowserStack MCP Server

• Open a mobile app on a specific device (e.g., iPhone 15 Pro Max) to debug crashes.
• Test a locally hosted website on Edge or Chrome without manual setup.
• Automatically analyze, diagnose, and fix broken automated test scripts.
• Create test management projects, add test cases, and run test suites.
• Run accessibility scans on local or production sites and get AI-suggested fixes.

FAQ from BrowserStack MCP Server

What are the runtime requirements?

Node.js version >=18.0 is required (recommended v22.15.0 LTS). You also need a BrowserStack account with a purchased plan (or free open-source plan).

How do I connect my BrowserStack account?

Provide your username and access_key from your BrowserStack Account Settings as environment variables (BROWSERSTACK_USERNAME and BROWSERSTACK_ACCESS_KEY) in the MCP server configuration.

Does the server automatically apply code changes?

No. When fetching Root Cause Analysis for a test failure, the server returns the suggested fix as a proposal only. It never applies code changes automatically; your assistant must present the suggestion and wait for explicit approval.

What IDEs or clients are supported?

Supported clients include VS Code (Copilot Agent Mode), Cursor, Claude Desktop, and Cline. One-click setup is available for VS Code and Cursor.

How many tools does the server provide?

As of now, the server supports 20 tools covering test management, manual web/app testing, automated testing, accessibility scanning, and AI agent capabilities.