* [Midscene.js v1.8.7](https://github.com/web-infra-dev/midscene) – AI-driven tool for operating web pages and extracting data using natural language commands.

* [native-devtools-mcp v0.10.1](https://github.com/sh3ll3x3c/native-devtools-mcp) – MCP server enabling screenshots, OCR, input simulation, window management, and Chrome/Electron (CDP) browser automation on macOS, Windows, and Android.

* [computer-use-linux v0.2.4](https://github.com/agent-sh/computer-use-linux) – Linux desktop control server and CLI for MCP hosts, offering AT-SPI accessibility, compositor-aware window targeting, screenshots, and input.

* [Clawd Cursor v0.9.9](https://github.com/AmrDab/clawdcursor) – OS- and model-agnostic skill that enables AI agents to control mouse, keyboard, read screens, and interact with GUIs on Windows, macOS, and Linux.