take_screenshot
Capture app window as PNG
How to use it
take_screenshot is exposed by the Desktop Pilot MCP MCP server. Add the server to your MCP client (Claude Desktop, Cursor, Windsurf and others), and the take_screenshot tool becomes available to the model automatically. See the full listing for setup details and every tool this server provides.
Install Desktop Pilot MCP
npx -y winapp-mcpOther tools in Desktop Pilot MCP (39)
Draw red bounding boxes around elements
Attach to a running process by name
Attach to a running process by PID
Click at absolute screen coordinates
Click a UI element by AutomationId or name
Click in a specific window by handle
Close a tracked application
Double-click a UI element
Drag from one element to another
Fast boolean check — does this element exist?
Expand, collapse, or toggle tree/menu items
Fill multiple form fields in one call
List all matching elements with indices
Search for elements with filters (type, id, name)
Fuzzy search — tolerates typos and partial names
Read all editable field values at once
Get bounding rectangle (screen coordinates)
Get the currently focused element's info
Get a UI tree snapshot (like browser DOM inspection)
Hash the UI tree to detect changes
Invoke via UIA InvokePattern/TogglePattern
Launch a Windows app by executable path
List all currently tracked applications
List all visible desktop windows
List all windows of a tracked app
Press a single key (RETURN, TAB, ESCAPE, etc.)
Press a key combination (Ctrl+S, Alt+F4)
Read detailed properties of a UI element
Read properties by index (from findallelements)
Emergency: release all stuck modifier keys
Right-click (open context menu)
Pixel-diff two screenshots, highlight changes
Select a ComboBox/dropdown option in one call
Set text in a specific window by handle
Screenshot with auto-resize for LLM token budgets
Type text into a text field
Wait for a property to reach a value
Wait for an element to appear
Wait for window to be ready for input