Skip to content

Conversation

@kxrm
Copy link
Owner

@kxrm kxrm commented Dec 23, 2025

This adds a floating bubble chat interface for interactive Claude conversations, along with several improvements to the vision tools.

Main changes:

  • New bubble.sh tool with a native macOS GUI for displaying messages, pointing at screen elements, and capturing user responses
  • Element detection for icons and buttons without text labels (focus-based drag operations)
  • --near flag for OCR disambiguation when multiple text matches exist
  • Drag operations between coordinates, text, and detected elements
  • --point-at for positioning bubbles relative to screen elements
  • Image extraction from --read-page output
  • Smart image resizing to 1568px for API compatibility
  • --preview flag on screenshot.sh to verify click coordinates before executing

The bubble interface uses a lightweight AppKit window that stays on top and can point arrows at specific UI elements, which is useful for guiding users through multi-step tasks.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants