name: virtual-desktop-browser description: Launch Chromium in non-headless mode inside Xvfb virtual display (fixed 1200x720x24) and automate with human-like mouse/keyboard/screenshot operations. Use for bot-resistant sites like Xiaohongshu and X/Twitter where GUI simulation is required.

Virtual Desktop Browser Skill

This skill provides a deterministic GUI automation runtime:

• Xvfb virtual display (1200x720x24)
• Chromium non-headless browser
• PyAutoGUI control (click/type/hotkey/scroll)

Required system packages

apt-get update
apt-get install -y xvfb chromium-browser \
  libnss3 libgconf-2-4 libxss1 libasound2 \
  libatk1.0-0 libatk-bridge2.0-0 libcups2 \
  libdrm2 libgbm1 libgtk-3-0 libxshmfence1 x11-utils

Python dependencies

pip install -r requirements.txt

Tool surface

All tools return JSON dicts.

• browser_start(url=None, display=None) — start Xvfb + Chromium
• browser_stop() — close processes
• browser_snapshot(region=None) — screenshot as Base64 PNG
• browser_click(x, y, button='left', clicks=1, duration=0.5) — move + click
• browser_type(text, interval=0.05, wpm=None) — type at focus
• browser_hotkey(keys, interval=0.05) — key combos
• browser_scroll(clicks=1, direction='vertical', x=None, y=None) — mouse wheel
• browser_find_image(image_path, confidence=0.8) — template matching (OpenCV)
• browser_get_pixel_color(x, y) — read RGB
• browser_activate_window(title_substring) — focus window by title

Notes

• Browser lifecycle is manual: start once, multi-step flow, then stop.
• Display auto-assigns (:99..:199) if not provided.
• Failsafe: moving mouse to bottom-right corner triggers pyautogui abort.

Language

See docs/ for translations:

• English
• Español
• العربية