v0.6.5

• AI stream ready — natively supports AI streaming output; feed a markdown list in, get a real-time mind map out
• Built-in AI generation — connect any OpenAI-compatible API to generate mind maps from natural language; supports text, image, and PDF attachments
• Pure SVG rendering — no canvas, no external layout engines, razor-sharp at any zoom level
• iOS-style UI — frosted glass controls, rounded corners, smooth animations, clean and polished design
• Plugin system — 7 built-in plugins for extended syntax (dotted lines, folding, multi-line, tags, cross-links, LaTeX, frontmatter); fully extensible
• Inline formatting — bold, italic, code, strikethrough, ==highlight==, and links inside nodes
• Task status — [x] done, [ ] todo, [-] in-progress checkboxes
• Remarks — multi-line remarks attached to nodes via > syntax
• Text editing mode — toggle between visual mind map and plain-text markdown editing
• Full-screen mode — expand the component to fill the viewport
• LaTeX math — render $...$ inline and $$...$$ display formulas (requires KaTeX)
• Cross-links — draw edges between arbitrary nodes via {#anchor} / -> {#target}
• Lightweight Viewer — a standalone read-only component (MindMapViewer) with ~48% smaller bundle; import via @xiangfa/mindmap/viewer for minimal footprint
• Readonly mode — display-only with pan/zoom/select but no editing; ideal for presentations and embeds
• Multiple root nodes — build separate trees on the same canvas
• Drag & drop — reorder siblings by dragging; drag root's children across the center line to rebalance sides
• Keyboard shortcuts — Enter to create, Delete to remove, Cmd+C/V to copy/paste, Shift+ shortcuts for zoom & layout
• Markdown I/O — feed a markdown list in, get a mind map out (great for AI streaming)
• i18n — auto-detects browser language; built-in Chinese and English, fully customizable via props
• Dark mode — auto-detects prefers-color-scheme, or set light / dark explicitly
• Export — SVG, high-DPI PNG, and Markdown export out of the box
• Import — paste JSON or markdown data via the context menu import dialog
• Context menu — right-click to add root nodes, import data, export, or change layout
• Layout modes — left, right, or balanced (both) layout directions
• Mobile optimized — full touch support with single-finger pan/drag and two-finger pinch-to-zoom centered on content
• Toolbar control — show/hide zoom controls via the toolbar prop
• Tiny footprint — zero runtime dependencies beyond React

---