mcp-operator

mcp-operator

2.7
browser_automation

The MCP Browser Operator is an MCP server specialized in automating web browsers. It enables management and interaction with web content through JSON-RPC, with features like web page auditing and persistent browser states for seamless session handling.

MCP Browser Operator

A Model Control Protocol (MCP) server designed for browser automation, enabling the control of a web browser and interaction with web pages via a standardized JSON-RPC interface.

Features

  • Browser Management: Handle browser instances with create, navigate, operate, and close functions.
  • Job Management: Track browser operations using job IDs.
  • Web Interaction: Use OpenAI's Computer Use API for interactive tasks.
  • Browser Tools: Access console logs, network activity, and perform page audits for accessibility and performance.
  • Auditing: Conduct audits for SEO and other aspects.

Requirements

  • Python 3.11+
  • Playwright
  • OpenAI API key

Installation

  • Clone the repository and install dependencies and browsers.
  • Set up OpenAI API key.

Usage

  • Start the server which listens for JSON-RPC requests and manages jobs asynchronously.

Core Methods

  • Browser Management: Create, navigate, operate, and close browser instances.
  • Job Management: Get job status and list jobs.
  • User Notes: Add notes related to browser operations.

Asynchronous Workflow Pattern

Relies on job-based asynchronous operations to avoid client timeouts.

Persistent Browser State

Maintains state like cookies and local storage between sessions, supporting multiple independent browser states.