What you get

Features

Everything SiriusMarkdown renders, and what readers can do with it: stream, resize, select, copy, search, export, and use it with assistive technologies.

Built for streaming

Chat replies render as they arrive. Useful prepared prefixes publish while input continues, and supported native math and diagram children appear inside unfinished containers. Completed regions keep their identity while the unfinished tail updates. You can put your own views, such as tool results or attachments, between messages.

Prepared for resizing

Drag a window edge, rotate an iPad, or open a sidebar. Layout reuses prepared content at the new width.

Text, structure, and media

  • Paragraphs, headings, block quotes, and horizontal rules
  • Ordered, unordered, nested, and task lists
  • Code blocks with syntax highlighting, plus Copy, Export, and Collapse buttons
  • Tables with alignment and merged cells. Wide tables scroll sideways instead of squeezing.
  • Links and autolinks with fallback symbols and available site icons
  • Images, shown only where you allow them
  • A safe subset of HTML, rendered as native text and tables rather than shown as tags
  • Complete subscript and superscript glyphs, including H₂O, in compact prose and table cells

Math

Inline and display LaTeX is typeset by default, including in table cells. It handles the notation textbooks and language models actually write, such as \dfrac, \operatorname, align*, cases, and \underbrace, plus slashed expressions such as \not{D} and native labeled arrows. Readers can select an equation and copy it as LaTeX, and VoiceOver reads its structure. See how it compares with other engines on the comparison page.

Diagrams

Mermaid blocks render as native diagrams, in light and dark mode. Sixteen types are supported: flowchart, sequence, class, entity-relationship, state, XY chart, quadrant, pie, journey, timeline, mindmap, Gantt, Sankey, GitGraph, radar, and treemap. Readers can zoom and pan, switch to the source, and export a diagram as PDF or SVG. If a block uses syntax that isn't supported, the reader sees its source rather than a broken drawing.

Select, copy, and find

Selection works across paragraphs, lists, tables, code, and math, with the mouse and keyboard on Mac and with the standard handles and edit menu on iPhone and iPad. Copying gives you the exact Markdown and plain text. Complete selections can also provide HTML, with RTF on macOS. Find searches the whole document, and links to headings work.

On Mac, documents print and export to PDF with selectable text, tables that repeat their headers across pages, working links, and typeset math. Diagrams export on their own as PDF or SVG.

Reader-page export on macOS and iPadOS captures the chosen presentation as raster PDF pages. Optional columns and compact-table settings help fit dense documents; raster pages have fixed resolution and no selectable text or link annotations.

Editing and media

Hosts can opt into formatting, image resizing, and table commands while retaining ownership of source edits and Undo. Local video attachments use prepared posters and native playback controls, and begin playing only when the reader asks. The SDK also supports a host-owned macOS Quick Look preview extension.

Accessible

Links are standard accessibility actions. Native macOS tables expose row and column relationships. Equations and diagrams carry spoken descriptions. Lists and quotes read item by item.

Safe by default

Only web, email, and relative links work out of the box. Public HTTPS pictures load anonymously by default; hosts can opt out. Link icons are fetched anonymously from public HTTPS sites only, and you can switch them off. Scripts, embedded pages, and other active HTML are removed, never run. Your app controls link and resource access through replaceable policies.

Images and text flow

Figures support width, alignment, captions, and descriptions. Markdown figures can wrap prose on the left or right, with a full-width layout when there is too little room beside the image. Inspect the wide and narrow examples. Native HTML figures share the image and caption pipeline; arbitrary CSS layout is not supported.

Light, dark, and your own style

Two built-in styles cover reading documents and compact chat. You can change fonts, spacing, colors, and the look of each kind of block.

Where it runs

macOS 13, iOS/iPadOS 16, and visionOS 1, with Swift 6.3 or later. Math and diagrams render natively on these platforms.

For product selection and integration, read the SDK guide.

Native captureOpen original ↗
Native document capture