About Playwright Cloud Browsers
Playwright Cloud Browsers is a managed browser MCP service for AI agents, supporting navigation, inspection, and interaction through remote sessions. It uses Streamable HTTP and requires no local Playwright or browser installation; the MCP service is in preview.
Key features
- 01Create and close managed browser sessions
- 02Navigate to and inspect pages
- 03Interact with page elements
- 04Connect over Streamable HTTP
- 05Authenticate with Microsoft Entra ID
Use cases
- 01Ask an agent to open a page and report its main heading
- 02Inspect pages in a cloud browser
- 03Have an agent interact with web elements
- 04Run browser tasks without a local browser
MCP support
The remote MCP service gives AI agents managed browser sessions for navigation, page inspection, and interaction over Streamable HTTP. It is currently in preview.
Configuration
Remote servers need no local install. Pick your client, then copy the config or install in one click.
https://{region}.mcp.playwright.microsoft.com/playwrightworkspaces/{workspaceId}/mcpclaude mcp add --transport http com-microsoft-playwright-cloud-browsers https://{region}.mcp.playwright.microsoft.com/playwrightworkspaces/{workspaceId}/mcp{"mcpServers":{"com-microsoft-playwright-cloud-browsers":{"type":"http","url":"https://{region}.mcp.playwright.microsoft.com/playwrightworkspaces/{workspaceId}/mcp"}}}Open in Cursor{"servers":{"com-microsoft-playwright-cloud-browsers":{"type":"http","url":"https://{region}.mcp.playwright.microsoft.com/playwrightworkspaces/{workspaceId}/mcp"}}}Open in VS Code{"mcpServers":{"com-microsoft-playwright-cloud-browsers":{"type":"http","url":"https://{region}.mcp.playwright.microsoft.com/playwrightworkspaces/{workspaceId}/mcp"}}}{"mcpServers":{"com-microsoft-playwright-cloud-browsers":{"serverUrl":"https://{region}.mcp.playwright.microsoft.com/playwrightworkspaces/{workspaceId}/mcp"}}}Run commands in a terminal; paste JSON into the client’s MCP config and restart. Servers that need sign-in will prompt on first connect.
Tools
Details
From the project READMEPlaywright Workspaces
✅ Playwright Workspaces is now Generally Available!
Playwright Workspaces is a fully managed service for end-to-end web testing, built on top of Playwright. With Playwright Workspaces, you can significantly speed up test execution by running tests in parallel using managed cloud-based browers. The service takes care of the underlying infrastructure and complexity, allowing you to validate your web app across major browsers on both Linux and Windows—so you can ship features faster and with greater confidence.
👉 Get started with the Quickstart
Documentation
The full documentation is now available on learn.microsoft.com
There you’ll learn how to:
- Run and scale Playwright tests in the cloud
- Optimize test suite performance and reliability
- Test across multiple browsers and operating systems
- Set up secure, region-aware test environments
- Integrate with CI/CD
Playwright Cloud Browsers remote MCP (preview)
The Playwright Cloud Browsers remote MCP server gives AI agents managed cloud browsers over Streamable HTTP. Agents can create a browser session, navigate and inspect pages, interact with elements, and close the session without installing Playwright or a browser locally.
The remote MCP server is in preview, separately from the generally available Playwright Workspaces service. Preview features do not have a service-level agreement and are not recommended for production workloads.
Prerequisites and authentication
You need an Azure subscription, a Playwright workspace with access to the remote MCP preview, its region and workspace ID, and an identity with permission to use browser sessions in that workspace.
Use the endpoint for your workspace's region:
https://<region>.mcp.playwright.microsoft.com/playwrightworkspaces/<workspace-id>/mcp
Replace both placeholders with your workspace values. Use Microsoft Entra ID with a client that supports OAuth discovery. The server publishes protected-resource metadata; your signed-in identity must also have the required workspace RBAC access. See the authentication and setup guide for permissions and supported compatibility options. Never commit access tokens or paste them into prompts.
Connect from VS Code or the GitHub Copilot app
- VS Code: Run MCP: Add Server from the Command Palette, select HTTP, enter the resolved workspace endpoint, and choose a configuration destination. Follow the server trust and Microsoft Entra sign-in prompts. See Add MCP servers in VS Code.
- GitHub Copilot app: Open Customize > MCP, add a custom server using the resolved workspace endpoint, and follow the supported authentication prompts. See Configuring MCP servers.
With the server connected, ask the agent to create a browser session, open
https://example.com, report the page's main heading, and close the session,
including on failure. Inspect the actual MCP tool calls and outputs to verify
the connection. See the service guide for session
limits, supported tools, and troubleshooting authentication, timeouts, and
expired sessions. Report service issues through this repository's
issue tracker.
Registry metadata
The server is published as com.microsoft/playwright-cloud-browsers in the
Official MCP Registry.
The root server.json is a public copy of the published
metadata. Publication and metadata updates are managed by the service team;
changing this copy does not publish an update or establish a listing in
downstream client catalogs.
Samples
Explore sample projects for different test runners:
- Playwright Test with JavaScript/TypeScript
- NUnit with C#
- Python
- Playwright Library projects
- Foundry Prompt Agents using Playwright Workspaces Remote MCP
- Foundry Hosted Agents using Playwright Workspaces Remote MCP
- App Verification Skills (verify skill)
Migrating from Microsoft Playwright Testing
- See the Migration Guide and Troubleshooting Tips
Issues
To report a bug, request a feature, or ask for a quote limit increase, please use the Issues section of this repository.
More Resources
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.