Skip to main content

A powerful MCP server for PyAutoGUI

Project description

MCP Server for PyAutoGUI

License: GPL v3 PyPI - Version PyPI - Python Version PyPI - Status

mcp-pyautogui implements a Model Context Protocol (MCP) server for PyAutoGUI. It provides control of mouse, keyboard, and screen interactions, allowing AI systems to automate tasks on a host machine.

The server runs over stdio transport and provides tools for clicking, typing, taking screenshots, and more, with a focus on simplicity and reliability. It’s designed for developers and AI enthusiasts looking to integrate desktop automation into their workflows.

Features

  • Control mouse movements, clicks, and drags.
  • Simulate keyboard input (single keys, text, hotkeys).
  • Capture screenshots and retrieve mouse position.
  • Get the current operating system for platform-specific logic.
  • Consistent error handling with boolean success indicators (where applicable).

Prerequisites

  • Python 3.11+

You'll also have to grant the relevant permissions.

Installation

Use pip to install the server.

pip install mcp-pyautogui

You can then run the following comand to determine the location at which the server was installed:

which mcp-pyautogui

# Sample output:
# /Users/bob/miniconda3/envs/desk/bin/mcp-pyautogui

Use that location when you're adding mcp-pyautogui to your claude_desktop_config.json file. Here's a sample configuration:

{
  "mcpServers": {
    "mcp-pyautogui": {
      "command": "/Users/bob/miniconda3/envs/desk/bin/mcp-pyautogui",
      "args": [
          ""
      ]
    }
  }
}

Usage

If you followed the above installation steps, you don't need to do anything else. Just use Claude Desktop and you should see all the tools this MCP server offers automatically. You could test this server by saying something like "can you click at (200,200) ?".

The server runs over stdio transport, meaning it communicates through standard input/output. To use it, connect an MCP-compatible client (e.g., an AI assistant or custom script) that can send JSON-RPC commands and receive responses.

Example Commands

  • Click at (100, 200): click(100, 200) → Returns true if successful.
  • Type "hello": type_text("hello") → Returns true.
  • Take a screenshot: take_screenshot("screenshot.png") → Saves to screenshot.png.
  • Get mouse position: get_mouse_position() → Returns (500, 300) (example).
  • Copy text: hotkey("ctrl c") → Returns true.
  • Get current operating system: get_os() → Returns "Windows" (if on Windows).

Full List of Available Tools

  • click(x, y): Left-click at coordinates (x, y).
    • Parameters: x: int, y: int
    • Return Type: bool
  • right_click(x, y): Right-click at coordinates (x, y).
    • Parameters: x: int, y: int
    • Return Type: bool
  • move_to(x, y): Move mouse to coordinates (x, y) over 3 seconds.
    • Parameters: x: int, y: int
    • Return Type: bool
  • drag_to(x, y, duration): Drag mouse to (x, y) over specified duration (default 1.0s).
    • Parameters: x: int, y: int, duration: float
    • Return Type: bool
  • type_text(text): Type the given text.
    • Parameters: text: str
    • Return Type: bool
  • press_key(key): Press and release a single key (e.g., 'enter', 'space', 'a').
    • Parameters: key: str
    • Return Type: bool
  • hotkey(keys): Press multiple keys together (e.g., 'ctrl c'). Space-separated string.
    • Parameters: keys: str
    • Return Type: bool
  • scroll(amount): Scroll up (positive) or down (negative) by amount.
    • Parameters: amount: int
    • Return Type: bool
  • take_screenshot(filename): Save a screenshot to the specified file (e.g., 'screen.png').
    • Parameters: filename: str
    • Return Type: bool
  • get_mouse_position(): Get current mouse coordinates as (x, y).
    • Parameters: None
    • Return Type: str
  • get_os(): Get the current OS name (e.g., 'Windows', 'macOS', 'Linux').
    • Parameters: None
    • Return Type: str
  • double_click(x, y): Perform a double-click at the given (x, y) coordinates.
    • Parameters: x: int, y: int
    • Return Type: bool
  • get_screen_size(): Get the screen resolution as (width, height).
    • Parameters: None
    • Return Type: str
  • pixel_color(x, y): Get the RGB color of the pixel at (x, y). Returns (r, g, b).
    • Parameters: x: int, y: int
    • Return Type: str

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

mcp_pyautogui-0.0.5.tar.gz (16.9 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

mcp_pyautogui-0.0.5-py3-none-any.whl (17.2 kB view details)

Uploaded Python 3

File details

Details for the file mcp_pyautogui-0.0.5.tar.gz.

File metadata

  • Download URL: mcp_pyautogui-0.0.5.tar.gz
  • Upload date:
  • Size: 16.9 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.12.11

File hashes

Hashes for mcp_pyautogui-0.0.5.tar.gz
Algorithm Hash digest
SHA256 262952a30ef66190ac929843ef87876fbe653317376d58984d8ea3b6299b3677
MD5 85a6457ceb24b2faf66cd96680f78eaa
BLAKE2b-256 f832ee17cabd964eaa8b03cf283adc129488995afe64b5955e6875b10badb142

See more details on using hashes here.

File details

Details for the file mcp_pyautogui-0.0.5-py3-none-any.whl.

File metadata

  • Download URL: mcp_pyautogui-0.0.5-py3-none-any.whl
  • Upload date:
  • Size: 17.2 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.12.11

File hashes

Hashes for mcp_pyautogui-0.0.5-py3-none-any.whl
Algorithm Hash digest
SHA256 9e46713f32bf2c2662d05b4b40ffc4c076db492ea317db3d18fba9ef2296a0ba
MD5 efa9e680814aef724302e26ea70a10f6
BLAKE2b-256 abe8a1807c866a8bbfdec92e6f404be163792016510808c79dd27c9e054e5d2f

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page