MCP server for browser and desktop automation (RPA) with OCR, image recognition and real mouse/keyboard input. Works with Claude Code, Claude Desktop, Cursor, Windsurf, VS Code, LM Studio and any other MCP client.
Click at the specified coordinates on the screen
Create a new macro with specified name and script or command content
Search for an image on the screen using vision-based image recognition
Get the full definition of a specific macro
List all available macros
Extract text from an image or screenshot using OCR
Run a macro by name and optionally pass parameters and screenshot context
Take a screenshot of the current browser or desktop view
Set or update a macro definition with script content or command table
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 21 | 2026-07-28+ | v2 |
Type text at the current cursor position