desktop-computer-automation
Vision-driven desktop automation for native apps using natural language commands and screenshots.
- Controls macOS, Windows, and Linux desktops entirely from visual input; no DOM or accessibility labels required
- Operates synchronously with a screenshot-analyze-act loop: connect, observe screen state, execute high-level actions via natural language prompts, then disconnect
- Requires a vision-capable AI model (Gemini, Qwen, Doubao, or similar) configured via environment variables; supports multiple model providers and OpenRouter
- Takes over the user's mouse and keyboard during execution; best suited for desktop-native apps (Electron, Qt, native UIs) that cannot run headless; web apps should use Browser Automation instead
Desktop Computer Automation
CRITICAL RULES — VIOLATIONS WILL BREAK THE WORKFLOW:
- Never run midscene commands in the background. Each command must run synchronously so you can read its output (especially screenshots) before deciding the next action. Background execution breaks the screenshot-analyze-act loop.
- Run only one midscene command at a time. Wait for the previous command to finish, read the screenshot, then decide the next action. Never chain multiple commands together.
- Allow enough time for each command to complete. Midscene commands involve AI inference and screen interaction, which can take longer than typical shell commands. A typical command needs about 1 minute; complex
actcommands may need even longer.- Always report task results before finishing. After completing the automation task, you MUST proactively summarize the results to the user — including key data found, actions completed, screenshots taken, and any relevant findings. Never silently end after the last automation step; the user expects a complete response in a single interaction.
- Only minimize windows, never close them unless explicitly asked. When you need to dismiss or get a window out of the way, minimize it instead of closing it. Do not close any app or window unless the user explicitly asks you to do so.
Control your desktop (macOS, Windows, Linux) using npx -y @midscene/computer@1. Each CLI command maps directly to an MCP tool — you (the AI agent) act as the brain, deciding which actions to take based on screenshots.
What act Can Do
Inside a single act call on desktop, Midscene can move the mouse, click, double-click, right-click, drag items, type or clear text, scroll, press single keys or keyboard shortcuts, and work through multi-step interactions on whatever is visible on the selected display.
Prerequisites
Midscene requires models with strong visual grounding capabilities. The following environment variables must be configured — either as system environment variables or in a .env file in the current working directory (Midscene loads .env automatically):
More from web-infra-dev/midscene-skills
browser-automation
|
2.8Kandroid-device-automation
>
1.6Kios-device-automation
|
1.4Kharmonyos-device-automation
>
1.0Kvitest-midscene-e2e
Enhances Vitest with Midscene for AI-powered UI testing across Web (Playwright), Android (ADB), and iOS (WDA). Scaffolds new projects, converts existing projects, and creates/updates/debugs/runs E2E tests using natural-language UI interactions. Triggers: write test, add test, create test, update test, fix test, debug test, run test, e2e test, midscene test, new project, convert project, init project, 写测试, 加测试, 创建测试, 更新测试, 修复测试, 调试测试, 运行测试, 新建工程, 转化工程.
838chrome-bridge-automation
|
665