# Streamdown

> A headless React renderer for LLM markdown: caches finished blocks, reveals text smoothly, completes half-typed syntax.

Each entry links to a markdown reference page assembled from its MDX source; the Docs URL is the rendered page with live demos.

Agent skill entry point: [skills.md](https://streamdown.lobehub.com/skills.md).

## Guides

- [Code Fences](https://streamdown.lobehub.com/skills/content/code-fences.md): Render open and closed code fences while streaming, plug in your own highlighter and diagrams, and detect an open fence. Outline: Open fences, Highlighting with components, findOpenFenceLanguage. Docs: https://streamdown.lobehub.com/content/code-fences
- [Incomplete Markdown](https://streamdown.lobehub.com/skills/features/incomplete-markdown.md): How Streamdown completes unclosed markdown on the open tail with remend, so partial emphasis, links and code render formatted. Outline: How it works, Options. Docs: https://streamdown.lobehub.com/features/incomplete-markdown
- [Introduction](https://streamdown.lobehub.com/skills/guide/introduction.md): Why markdown streamed from an LLM trips up ordinary renderers, and how Streamdown keeps it readable while it arrives. Outline: Why streaming breaks markdown, What Streamdown does about it, What it leaves to you. Docs: https://streamdown.lobehub.com/guide/introduction
- [Block Memoization](https://streamdown.lobehub.com/skills/features/blocks.md): How Streamdown splits streaming markdown into top-level blocks, freezes finished ones and re-renders only the open tail. Outline: Block splitting, Frozen prefix, Commit pacing, Block states, Limitation: cross-block references. Docs: https://streamdown.lobehub.com/features/blocks
- [Getting Started](https://streamdown.lobehub.com/skills/guide/getting-started.md): Install @lobehub/streamdown, render your first streaming message, then add GFM tables and math. Outline: Requirements, Installation, Render a stream, Add GFM and math, Style it, Next steps. Docs: https://streamdown.lobehub.com/guide/getting-started
- [LaTeX](https://streamdown.lobehub.com/skills/content/latex.md): Render math while it streams with remark-math and rehype-katex, hold incomplete formulas with latexGuard, and clean model output with preprocessLaTeX. Outline: Setup, latexGuard, preprocessLaTeX. Docs: https://streamdown.lobehub.com/content/latex
- [Components & Plugins](https://streamdown.lobehub.com/skills/content/components.md): Style Streamdown's headless output, swap elements through react-markdown components, and add GFM, math, highlighting or diagrams with plugins. Outline: Override elements, Add capabilities, Plugins see one block at a time, Styling. Docs: https://streamdown.lobehub.com/content/components
- [Smooth Reveal](https://streamdown.lobehub.com/skills/features/smooth-reveal.md): The smoothing presets that turn bursty LLM chunks into a steady per-character reveal, and the hook behind them. Outline: Presets, useSmoothStreamContent. Docs: https://streamdown.lobehub.com/features/smooth-reveal
- [Usage](https://streamdown.lobehub.com/skills/guide/usage.md): Feed Streamdown from any token stream, use it with the AI SDK, and move over from react-markdown. Outline: From a fetch stream, With the AI SDK, Coming from react-markdown. Docs: https://streamdown.lobehub.com/guide/usage
- [Animation](https://streamdown.lobehub.com/skills/features/animation.md): How Streamdown fades newly revealed text in per character or per word with CSS, and how to style or disable it. Outline: How it works, granularity, Styles. Docs: https://streamdown.lobehub.com/features/animation

## src

- [Streamdown](https://streamdown.lobehub.com/skills/components/streamdown.md): Category: General. Streamdown renders partial markdown from an LLM stream with block caching, smooth per-character reveal and a LaTeX guard. It ships no stylesheet. Outline: Introduction, Basic Usage, Playground, API, Lower-level API, Limitations. Docs: https://streamdown.lobehub.com/components/streamdown
- [Streamdown Profiler](https://streamdown.lobehub.com/skills/components/profiler.md): Category: General. Measure Streamdown's root and block commit cost, reveal frames and input chunks with createStreamdownProfiler and StreamdownProfilerProvider from @lobehub/streamdown/profiler. Outline: Introduction, High-TPS stream, createStreamdownProfiler, Snapshot, useStreamdownProfiler. Docs: https://streamdown.lobehub.com/components/profiler

## Optional

- [Changelog](https://streamdown.lobehub.com/changelog): New updates and improvements to @lobehub/streamdown Docs: https://streamdown.lobehub.com/changelog
- [Streamdown](https://streamdown.lobehub.com/): A headless React renderer for LLM markdown: caches finished blocks, reveals text smoothly, completes half-typed syntax. Docs: https://streamdown.lobehub.com/