skills.rest
web-infra-devweb-infra-devOfficial·5 Agent Skills Included

midscene-skills

Vision-driven UI automation and testing across all platforms

Automates browsers, desktops, Android, iOS, and HarmonyOS devices using natural language commands driven by screenshots. Eliminates fragile selectors, manual clicking, and platform-specific scripting when testing or controlling apps. Scaffolds and runs Vitest end-to-end tests for Web, Android, and iOS from plain-English instructions.
286|22
↗
npx skills add web-infra-dev/midscene-skills --all -g -y

All Skills in This Repository (5)

Pure Emerald Level Indicators
📦 In Repo
web-infra-devweb-infra-dev

iOS Device Automation

Automate iOS device UI interactions via natural language commands through WebDriverAgent.

Official
Advanced
📦 In Repo
web-infra-devweb-infra-dev

Android Device Automation

Automates Android device actions via natural language and screenshot analysis.

Official
Intermediate
📦 In Repo
web-infra-devweb-infra-dev

Browser Automation

Automate web browser interactions using visual cues from screenshots.

Official
Intermediate
📦 In Repo
web-infra-devweb-infra-dev

Desktop Computer Automation

Automate desktop environments on macOS, Windows, and Linux via visual input.

Official
Advanced
📦 In Repo
web-infra-devweb-infra-dev

Chrome Bridge Automation

Automate Chrome browser tasks via screenshots while preserving login sessions.

Official
Intermediate

Frequently Asked Questions

FAQPage Schema
How to install Midscene Skills?▼

Run `npx skills add web-infra-dev/midscene-skills --all -g -y` in your terminal to install all platform skills globally. You also need to set a vision model API key in your environment variables.

Which platforms does Midscene support?▼

It covers web browsers via Puppeteer or CDP, desktop apps on macOS, Windows, and Linux, plus Android, iOS, and HarmonyOS devices. All platforms are controlled from screenshots with natural language.

How does Midscene automate UI without selectors?▼

It takes screenshots and uses a vision model to locate and interact with elements you describe in plain English. This makes automation resilient to layout and code changes.

Can Midscene write and run E2E tests?▼

Yes. The vitest-midscene-e2e skill scaffolds projects and creates, updates, debugs, and runs end-to-end tests for Web, Android, and iOS using natural-language test steps.

Do I need coding experience to use Midscene?▼

No. You describe tasks in plain English, such as opening an app or filling a form, and your agent executes the steps. Only a one-time model API key setup is required.

Related Repositories in Software Engineering

View All in Software Engineering→
openclawopenclaw
388.4k

openclaw

Run a personal AI assistant across your devices and chat apps

Official📦 120 Skills
obraobra
280.0k

superpowers

Gives coding agents a disciplined workflow from idea to merged code

Community📦 15 Skills
facebookfacebook
248.4k

react

AI agent skills for building, testing, and porting React core

Official📦 14 Skills