pov
Enables LLM agents to capture screenshots, control mouse/keyboard, and manage windows on desktop platforms, primarily Windows, via an MCP server.
README
pov - Point of View
Screenshot capture, mouse/keyboard control, and window management — as a CLI and MCP server for LLM vision agents.
Lets LLMs (Claude Desktop, Cursor, OpenCode) see and interact with the Windows desktop. Works on Windows, macOS, Linux (X11/Wayland), and WSL.
Install
# With uv
uv pip install pov-desktop
# With pip
pip install pov-desktop
To install the pov CLI as a standalone tool:
uv tool install pov-desktop
Or from source:
git clone https://github.com/empezarcero/pov.git
cd pov
uv sync
CLI Usage
# Screenshots
pov capture # Capture screenshot, print base64
pov capture --output shot.png # Save to file
pov capture --monitor 1 # Specific monitor
pov monitors # List available monitors
# Mouse
pov click # Left-click at current position
pov click 500 300 # Left-click at (500, 300)
pov click 500 300 --button right --clicks 2
pov move-mouse 800 600 # Move cursor to (800, 600)
pov scroll -3 # Scroll down 3 clicks
pov scroll 5 --x 400 --y 200 # Scroll up at position
pov cursor # Print current cursor position
# Keyboard
pov type "Hello, world!" # Type text literally
pov key enter # Press a key
pov key ctrl+c # Key combo
pov key alt+f4 # Another combo
# Windows
pov windows # List visible windows
pov processes # List windowed processes
pov foreground # Get the focused window
pov focus <hwnd> # Focus a window by handle
pov window-state <hwnd> minimize
pov move-window <hwnd> --x 100 --y 100
pov resize-window <hwnd> --width 1024 --height 768
pov close-window <hwnd> # Gracefully close a window
# MCP server
pov serve # stdio transport (default)
pov serve --transport sse --port 8000
MCP Server
The MCP server exposes 16 tools:
Screenshots
| Tool | Description |
|---|---|
screenshot |
Capture a screenshot and return it as an image |
list_monitors |
List available monitors and their geometry |
Mouse
| Tool | Description |
|---|---|
mouse_move |
Move the cursor to screen coordinates |
mouse_click |
Click at coordinates (or current position) |
mouse_scroll |
Scroll the mouse wheel |
get_cursor_position |
Get the current cursor position |
Keyboard
| Tool | Description |
|---|---|
keyboard_type |
Type a string of text literally |
keyboard_key |
Press a key or combo (e.g. ctrl+c, alt+f4) |
Window Management
| Tool | Description |
|---|---|
list_windows |
List all visible windows with handles, titles, positions |
focus_window |
Bring a window to the foreground |
set_window_state |
Minimize, maximize, restore, hide, or show a window |
move_window |
Move and/or resize a window |
resize_window |
Resize a window without moving it |
get_foreground_window |
Get info about the currently focused window |
close_window |
Gracefully close a window (WM_CLOSE) |
list_processes |
List running processes with visible windows |
Claude Desktop / Cursor / OpenCode
Add to your MCP config:
{
"mcpServers": {
"pov": {
"command": "uv",
"args": ["run", "--directory", "/path/to/pov", "pov", "serve"]
}
}
}
SSE Transport
{
"mcpServers": {
"pov": {
"url": "http://127.0.0.1:8000/sse"
}
}
}
Platform Support
| Platform | Backend | Notes |
|---|---|---|
| Windows | mss + ctypes |
Screenshots via mss, input/windows via Win32 |
| macOS | mss |
Screenshot capture only |
| Linux | mss |
Screenshot capture only (X11 and Wayland) |
| WSL | PowerShell + .NET/Win32 | Full support — captures the Windows desktop, controls mouse/keyboard, manages windows |
Development
# Install in development mode
uv sync
# Run the CLI
uv run pov --help
# Run the MCP server
uv run pov serve
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.