WebScout MCP

WebScout MCP is an automation-focused MCP server that helps you reverse engineer web applications, discover streaming endpoints, and capture network traffic. It provides browser automation, real-time traffic monitoring, and a developer-friendly toolset to analyze chat interfaces and streaming APIs, making it easier to identify how a web app communicates with its server endpoints.

How to use

To use WebScout MCP, add the server to your MCP client configuration so you can run it locally and coordinate browser-driven analysis with your existing MCP tooling. You can start a local MCP process and interact with the server to initialize a browser session, navigate to target pages, trigger chat interactions, and capture network traffic for analysis.

Interactive workflow and practical usage tips

• Create a new browser session for testing
• Navigate to a target URL such as a chat interface
• Trigger a message to start streaming responses
• Start and stop network capture to collect relevant data
• Review captured data to identify streaming patterns (SSE, WebSocket frames, chunked transfers)
• Take screenshots or retrieve page information for UI context

The most common workflow combines session management, navigation, interaction, and network capture. You can perform actions like filling forms, clicking elements, and waiting for dynamic content while the server captures and analyzes streaming data in real time.

MCP server configuration (stdio)

{
  "mcpServers": {
    "webscout-mcp": {
      "command": "npx",
      "args": ["-y", "webscout-mcp"]
    }
  }
}

Additional usage notes

If you prefer running directly from source, you can start the MCP server with a direct node invocation after installing dependencies. This is useful for development or debugging sessions where you want visible browser output or troubleshooting visibility.

Development mode tips

For debugging with a visible browser, run the server in development mode and set headless to false during session initialization. This helps you observe UI interactions and streaming behavior directly in the browser window.

Available tools

initializeSession

Create a new browser session for interactive operations by launching a controlled Playwright browser instance.

closeSession

Clean up browser resources and end a session, ensuring cookies and authentication state are properly released.

navigateToUrl

Navigate the active browser session to a specified URL within the same session context.

switchTab

Switch between open browser tabs in the current session to manage multiple targets.

takeScreenshot

Capture screenshots of the current viewport, full page, or specific elements for visual verification.

getCurrentPageInfo

Retrieve detailed information about the current page, including URL, title, and tab state.

clickElement

Interact with interactive elements by simulating clicks on buttons, links, or UI controls.

fillForm

Fill input fields and optionally submit forms as part of automated interactions.

waitForElement

Pause execution until a dynamic element appears or a timeout elapses.

startNetworkCapture

Begin monitoring network traffic with configurable filters and capture settings.

stopNetworkCapture

End the capture session and return the collected network data for analysis.

getNetworkCaptureStatus

Check the current status and statistics of an active network capture.

clearNetworkCapture

Remove all captured data without stopping the capture session.

reverseEngineerChat

Automated analysis of chat interfaces with streaming endpoint discovery to identify how messages are streamed.