React Grab Tool Drops
React Grab just launched — select any page element, AI changes it via Claude/Codex integration. Open-source tool hit 7,760 likes and 859k views, install with `npx react-grab@latest`. Built by @aidenybai for rapid UI prototyping.
React Grab works by tapping into React's internal Fiber architecture, the core engine that manages components. This allows it to access the hierarchy, props, and state of components on any public React site, essentially creating a blueprint of the user interface without needing access to the source code. This method provides a more accurate 1:1 reconstruction compared to AI approaches that rely on screenshots. The tool was created by developer Aiden Bai, who is also known for other popular open-source projects like Million.js and React Scan. Bai created Million.js at age 16 and his projects have collectively earned over 40,000 stars on GitHub. His focus is on improving developer experience and making web development more accessible through AI-powered tools. A key problem React Grab solves is providing precise context to AI coding assistants. Instead of an AI having to search through an entire codebase to find a specific element, developers can simply click on the UI component, and React Grab copies its exact location and context. This process can make AI agents like Claude Code up to three times faster and more accurate on UI tasks. The underlying technology that allows React Grab to access React's internals is a library called "bippy," also created by Aiden Bai. Bippy essentially mimics the behavior of React DevTools, allowing it to hook into the global variables that the official developer tools use to inspect the component tree. For safety, the tool is designed to run only in a development environment. An updated version, "React Grab for Agents," takes the concept a step further by allowing the tool to directly communicate with AI agents from the browser. Users can select an element, type a change they want to make, and the agent will directly edit the code files. This creates a more seamless workflow, removing the need to manually copy and paste context between the browser and the coding agent.