What problem does it solve? Manually copying UI element references from the browser into an AI coding session requires constant copy-paste handoffs. This Skill runs a hands-free loop where elements grabbed in the browser with React Grab are automatically delivered to the agent as actionable tasks. ## Core Features & Use Cases - Continuous grab watcher: Runs npx react-grab@latest pull in a blocking loop so every grabbed element arrives as JSON with source references (file:line, component name). - Mid-task redirection: New grabs captured while the agent is working supersede the current task, with non-blocking peeks between steps and background execution for slow commands. - Prompt-driven actions: When a grab includes a user-typed prompt, the agent executes that instruction directly against the referenced source file. - Use Case: A developer points at a misaligned button in their running app, grabs it with the comment "make this full width", and the agent immediately edits the correct component file without any manual context transfer. ## Quick Start Ask the agent to watch your React Grab grabs and act on each element you select in the browser until you say stop.