\n\n# Midscene Skills\n\n Vision-driven cross-platform automation \n\n- Natural-language driven UI control\n- Built on [Midscene.js](https://midscenejs.com)'s vision-based automation capabilities — Operates entirely from screenshots, making cross-platform support reliable\n- This repository contains Skills for the following platforms:\n - Browser (Puppeteer, headless Chrome; also supports CDP connection): [`skills/browser`](skills/browser)\n - Chrome Bridge (user's own Chrome browser): [`skills/chrome-bridge`](skills/chrome-bridge)\n - Desktop (macOS, Windows, Linux): [`skills/computer-automation`](skills/computer-automation)\n - Android (controlled via ADB): [`skills/android-automation`](skills/android-automation)\n - iOS (controlled via WebDriverAgent): [`skills/ios-automation`](skills/ios-automation)\n - HarmonyOS (controlled via HDC): [`skills/harmony-automation`](skills/harmony-automation)\n - Vitest + Midscene E2E (scaffold, convert, and manage AI-powered E2E tests for Web, Android, iOS): [`skills/vitest-midscene-e2e`](skills/vitest-midscene-e2e)\n\n\n## Safety Warning\n\n⚠️ AI-driven UI automation may produce unpredictable results since it can control EVERYTHING on the screen. Please evaluate the risks carefully before use.\n\n## Installation\n\nMake sure you have [Node.js](https://nodejs.org) installed.\n\nThen install the skills:\n\n```bash\n# General installation\nnpx skills add web-infra-dev/midscene-skills\n\n# Claude Code \nnpx skills add web-infra-dev/midscene-skills -a claude-code\n\n# OpenClaw\nnpx skills add web-infra-dev/midscene-skills -a openclaw\n```\n\n## Model Setup\n\nMidscene requires models with strong **visual grounding** capabilities (accurate UI element localization from screenshots). \nBecause of this, you need to prepare model access and configuration separately from skill installation.\n\nMake sure these environment variables are available in your system. You can also define them in a `.env` file in the current directory, and Midscene will load them automatically:\n\n```bash\nMIDSCENE_MODEL_API_KEY=\"your-api-key\"\nMIDSCENE_MODEL_NAME=\"model-name\"\nMIDSCENE_MODEL_BASE_URL=\"https://...\"\nMIDSCENE_MODEL_FAMILY=\"family-identifier\"\n```\n\nExample: Gemini (Gemini-3-Flash)\n\n```bash\nMIDSCENE_MODEL_API_KEY=\"your-google-api-key\"\nMIDSCENE_MODEL_NAME=\"gemini-3-flash\"\nMIDSCENE_MODEL_BASE_URL=\"https://generativelanguage.googleapis.com/v1beta/openai/\"\nMIDSCENE_MODEL_FAMILY=\"gemini\"\n```\n\nExample: Qwen3-VL\n\n```bash\nMIDSCENE_MODEL_API_KEY=\"your-openrouter-api-key\"\nMIDSCENE_MODEL_NAME=\"qwen/qwen3-vl-235b-a22b-instruct\"\nMIDSCENE_MODEL_BASE_URL=\"https://openrouter.ai/api/v1\"\nMIDSCENE_MODEL_FAMILY=\"qwen3-vl\"\n```\n\nExample: Doubao Seed 1.6\n\n```bash\nMIDSCENE_MODEL_API_KEY=\"your-doubao-api-key\"\nMIDSCENE_MODEL_NAME=\"doubao-seed-1-6-250615\"\nMIDSCENE_MODEL_BASE_URL=\"https://ark.cn-beijing.volces.com/api/v3\"\nMIDSCENE_MODEL_FAMILY=\"doubao-vision\"\n```\n\nCommonly used models: Doubao Seed 1.6, Qwen3-VL, Zhipu GLM-4.6V, Gemini-3-Pro, Gemini-3-Flash.\n\nModel setup docs:\n\n- Midscene model strategy: https://midscenejs.com/model-strategy\n- Qwen3-VL: https://midscenejs.com/model-common-config#qwen3-vl\n- Gemini-3-Flash: https://midscenejs.com/model-common-config#gemini-3-pro-and-gemini-3-flash\n- Doubao Seed 1.6: https://midscenejs.com/model-common-config\n\n\n## Use\n\nIn your chatbot or coding agent, you can say:\n\n```\nUse Midscene computer skill to open the Keynote app and create a new presentation.\n```\n\n```\nUse Midscene browser skill to open the Google search page and search for \"Midscene\".\n```\n\n## Issues\n\nFor bug reports, feature requests, and discussions, please visit the main Midscene repository: https://github.com/web-infra-dev/midscene/issues\n\n## License\n\nMIT\n",
"bytes": 3839,
"sha": "2b6d7dd3827a53582115760e96e7ade7b2233ee27cce6aaad62c95da36f49f2c",
"repo_slug": "web-infra-dev/midscene-skills",
"fonte": "repo",
"truncated": false,
"api": "https://agentalog.com/api/listings/skl_web_infra_dev_midscene_skills_chrome_bri_2625ee00/readme"
}