hive.browser-automation

What it does

All GCU browser tools drive a real Chrome instance through the Beeline extension and Chrome DevTools Protocol (CDP). That means clicks, keystrokes, and screenshots are processed by the actual browser's native hit testing, focus, and layout engines — not a synthetic event layer.

How it works

• Coordinates: tools operating on coordinates use fractions of the viewport (0..1 for both axes). Example mappings include browser_click_coordinate(x, y) and browser_hover_coordinate(x, y). Rect-returning tools return fractions and are converted to CSS pixels before dispatching to Chrome.
• Shadow DOM: browser_click_coordinate routes through CDP hit testing to traverse shadow roots; browser_type_focused inserts text into document.activeElement; browser_shadow_query and browser_get_rect support shadow-piercing with >>> syntax. The canonical click-type loop is to click with coordinates, verify focused_element from the response, then type if editable.
• Workflow: Start with browser_screenshot to inspect, identify target position, click via coordinates, verify focus, type into focused element with browser_type_focused(text=...), and verify via screenshots or browser_evaluate checks.
• Typing: ALWAYS click before typing into rich-text editors; typing uses Input.insertText by default for reliability in editors like Lexical, Draft.js, and ProseMirror. If per-keystroke dispatch is needed, use use_insert_text=False to fall back to keyDown/keyUp.
• Beforeunload: Use browser_evaluate to strip beforeunload handlers before navigating or closing tabs to avoid native confirm dialogs.

When to use it

• Use on shadow-heavy sites where shadow boundaries block selector-based interactions.
• When dealing with rich-text editors (X/Twitter, LinkedIn, Reddit, Gmail, Slack, Discord, Notion, Monaco) to ensure input events are properly registered.
• When needing reliable coordinates-based interaction that leverages CDP hit testing for accurate focus within complex layouts.

What it can touch

• Tools: browser_screenshot, browser_click_coordinate, browser_type_focused, browser_type, browser_shadow_query, browser_get_rect, browser_navigate, browser_wait_for_selector, browser_evaluate. The skill notes these as ways to drive a real Chrome instance via Beeline and CDP.

Caveats

• The workflow relies on CDP/Beeline and assumes a Chrome instance is available; shadow-heavy contexts require coordinate-based targeting rather than plain selector-based clicks.
• The typing reliability depends on default use_insert_text behavior; per-keystroke dispatch requires an option flag change.
• Some sites with dynamic SPAs may require navigation sleep to hydrate before interacting.