MCP Catalogs
← Home

MCP

by SelviAkdag·★ 0·Score 33

Windows-specific MCP server with browser automation and file management tools that actually work without Docker or WSL.

browser-automationfile-systemdeveloper-tools
0
Forks
0
Open issues
10 mo ago
Last commit
2d ago
Indexed

Overview

This repository provides production-ready MCP agents specifically configured for Windows environments. It offers two main agents: a Browser Agent with 24 Playwright tools for web automation, and a Filesystem Agent for file operations. The project addresses common Windows-specific issues that users typically encounter when setting up MCP servers, such as command errors, connection problems, and Chrome installation failures. The setup is designed to be straightforward with copy-paste configurations that work without modifications.

Try asking AI

After installing, here are 6 things you can ask your AI assistant:

you:Automating web browser tasks on Windows systems without virtualization
you:Managing files and directories through AI commands on Windows
you:Performing web searches and taking screenshots programmatically
you:What Windows-specific issues does this solve?
you:Do I need Docker or WSL to use this MCP server?
you:What prerequisites are needed?

When to choose this

Choose this if you're working on Windows and need MCP agents for browser automation or file management without virtualization overhead.

When NOT to choose this

Don't choose this if you're on Linux/Mac, need advanced features not covered by the provided agents, or prefer a more general MCP solution.

Tools this server exposes

5 tools extracted from the README
  • take_screenshot

    Takes a screenshot of a webpage and saves it as an image file.

  • navigate_to

    Navigates the browser to a specified URL.

  • search_google

    Performs a search on Google and returns the results.

  • list_files

    Lists files and directories in a specified folder.

  • create_file

    Creates a new file with specified content.

Note: Tool names inferred from example usage descriptions in the README. The README mentions 24 Playwright tools for browser automation but doesn't provide a complete list of specific tool names.

Comparable tools

mcp-server-playwrightmcp-server-filesystembrowserless-mcpnpx mcp-server-knowledge-base

Installation

Installation

  1. Clone the repository: git clone https://github.com/SelviAkdag/MCP.git
  2. Navigate to the project: cd MCP
  3. Create and activate virtual environment:

``bash python -m venv venv venv\Scripts\activate ``

  1. Install npx globally: npm install -g npx
  2. Install Hugging Face Hub with MCP support: pip install "huggingface_hub[mcp]>=0.32.0"
  3. Login to Hugging Face: huggingface-cli login
  4. Install MCP servers: npm install @playwright/test @modelcontextprotocol/server-filesystem
  5. Install browsers: npx playwright install (requires admin prompt)
  6. Run browser agent: tiny-agents run agents/agent_browser.json

For Claude Desktop, add to your config.json:

{
  "mcpServers": {
    "mcp-windows": {
      "command": "python",
      "args": ["-m", "tiny_agents", "run", "/path/to/MCP/agents/agent_browser.json"]
    }
  }
}

FAQ

What Windows-specific issues does this solve?
It solves common issues like 'KeyError: command', 'Connection closed', Chrome not found, and PowerShell execution policy blocks that Windows users typically encounter when setting up MCP servers.
Do I need Docker or WSL to use this MCP server?
No, this is designed to work directly on Windows without Docker or WSL. It runs on bare Windows with native support.
What prerequisites are needed?
Node.js 18+, Python 3.8+, and a Hugging Face account with API token. The setup guides you through creating these if needed.

Compare MCP with

MCP vs ultimate_mcp_serverMCP vs mcp-server-chartMCP vs everythingMCP vs filesystemMCP vs time
GitHub →

Last updated · Auto-generated from public README + GitHub signals.