mcp-chrome is a Chrome extension-based Model Context Protocol server that exposes browser functionality to AI assistants for automation and content analysis.
The tool solves the problem of integrating AI assistants with real browser environments by running as a Chrome extension rather than launching a separate browser process. This approach lets AI models control your existing Chrome instance with all your login states, configurations, and user habits intact. It provides over twenty tools for screenshots, network monitoring, interactive operations, bookmark management, and browsing history, along with semantic search capabilities powered by a built-in vector database and WebAssembly SIMD optimization for faster vector operations.
The project suits developers who want AI-assisted browser automation without the overhead of launching independent browser processes or re-authenticating. It works best for workflows where you need an AI assistant to interact with your actual browsing environment rather than a clean slate. Compared to Playwright-based MCP servers, the extension approach eliminates the need for separate browser binaries and inter-process communication overhead, preserves your existing browser environment and login states, and provides direct access to Chrome native APIs rather than being limited to Playwright's API surface.
The project is in early development stages with intensive ongoing work. The team has introduced a visual editor for Claude Code integration and continues adding stability improvements and new features.
