BrowseFleetMCP

Parallel browser control with isolated windows, real screenshots, and MCP support.

BrowseFleetMCP connects your MCP-compatible AI tools to your real Chromium/Chrome browser, so agents can work with web pages you already have access to while keeping everything local and inspectable. It is built for developers, operators, and automation-heavy users who want browser control that is more reliable than copy-pasting screenshots or manually describing pages to an agent. Once paired with the local BrowseFleetMCP server, the extension lets MCP clients create and control isolated browser sessions, inspect page structure, click elements, type into forms, capture screenshots, and navigate real websites through your existing browser environment. Why install BrowseFleetMCP? Most browser automation tools run in a fresh browser profile or a remote sandbox. That often breaks real workflows because supplier portals, dashboards, internal apps, and SaaS tools depend on your existing login state, cookies, browser permissions, and normal Chrome behavior. BrowseFleetMCP is designed for those practical workflows. With BrowseFleetMCP, agents can: - Connect to real browser tabs through a local MCP server - Create isolated browser windows for separate tasks - Work across multiple sessions without every agent fighting over one tab - Capture accessibility snapshots and rendered screenshots - Click, type, select, hover, drag, and press keys on real pages - Extract focused product/listing data from supplier and ecommerce pages - Recover from common page interaction issues with more robust click handling - Report browser, extension, session, and transport health for debugging The extension is especially useful for procurement, sourcing, QA, admin workflows, research, support operations, and any task where an AI assistant needs to inspect or operate a website that you can already access in your browser. Local-first by design BrowseFleetMCP does not provide a hosted browser service. The extension connects to a local BrowseFleetMCP server running on your machine, normally on 127.0.0.1. Your browser control path stays local: your MCP client talks to the local server, and the extension talks to that same local server. This design gives you more control over what is running, what is connected, and which browser sessions are available. The extension also exposes health and mismatch warnings so you can see whether the loaded extension, local server, and active broker are aligned. Important setup note This extension is one part of the BrowseFleetMCP system. To use it, you also need to install and run the BrowseFleetMCP local MCP server package. After installation, load or enable the extension, start the local MCP server, and connect a tab or create a new browser session from your MCP client. Install BrowseFleetMCP if you want AI agents to work with your real browser in a local, multi-session, inspectable way instead of relying on fragile screenshots, remote sandboxes, or one shared automation tab.