What problem does it solve? Rendering AI-generated Markdown in React apps is tricky when content streams in token by token: incomplete syntax breaks layouts, code blocks flicker, and raw HTML introduces security risks. This Skill guides you through setting up Streamdown, a drop-in react-markdown replacement built for streaming, with correct Tailwind configuration, plugins, and security hardening. ## Core Features & Use Cases - Streaming Markdown Rendering: Renders incomplete Markdown gracefully with remend auto-completion, block/circle carets, and per-block memoization for AI chat interfaces. - Plugin System: Add syntax highlighting (Shiki, 200+ languages), Mermaid diagrams, KaTeX math, and CJK text support via optional packages. - Security Controls: Configure link safety modals, protocol/prefix restrictions via rehype-harden, custom HTML tag whitelists, and URL transforms for AI-generated content. - Use Case: You are building a chat UI with the Vercel AI SDK and need assistant messages to render Markdown live as tokens arrive, with a blinking caret, copyable code blocks, and a confirmation modal before users open external links. ## Quick Start Ask the AI to set up Streamdown in your React chat component with the code plugin, Tailwind source configuration, and a block caret for streaming responses.