Semantic role: caption. Compiles to eHTML <caption>.

Registers:

• Canonical: <caption>

Semantic role: diagram. Compiles to eHTML <diagram>.

Registers:

• Canonical: <diagram>

<diagram mermaid>
graph LR
  A[Start] --> B{Decision}
  B -->|yes| C[OK]
  B -->|no| D[Stop]
</diagram>

graph LR
  A[Start] --> B{Decision}
  B -->|yes| C[OK]
  B -->|no| D[Stop]

The mermaid format word selects the Mermaid engine. The handler emits the same <pre class="mermaid" data-enscribe-dsl="mermaid"> contract as the legacy <mermaid> shorthand expands to.

<diagram abc>
X:1
T:Scale
K:C
CDEFGABc
</diagram>

X:1
T:Scale
K:C
CDEFGABc

The abc format word selects the abcjs engine.

Semantic role: figure. Compiles to eHTML <fig>.

Registers:

• Canonical: <fig>
• Alias: <figure>

The simplest case. The src kwarg generates the <img>; the pipe content generates the figcaption. The alt text defaults to the figcaption text when not specified explicitly. The eHTML element is HTML-native <figure> (not the custom-element <fig>) because the HTML rendering surface is the HTML5 native element while the enscribe canonical name follows JATS’s shorter <fig>.

<figure> is the authoring alias. The normalize-to-canonical gate rewrites the tagname to fig early; the rendered output is the same.

The id enables cross-referencing with <ref @elephant> (or the canonical <ref @fig:elephant> colon-prefix form). Numbered by default; the figcaption gets a “Figure N.” label span prepended.

<fig #revenue-table type=table |
<table>
  <tr><th>Quarter</th><th>Revenue</th></tr>
  <tr><td>Q1</td><td>$100M</td></tr>
  <tr><td>Q2</td><td>$120M</td></tr>
</table>
Quarterly revenue for fiscal year 2024.
>

A figure without src. The content (a table) is preserved as-is; the trailing line becomes the figcaption. Author convention is to put the caption text on its own line at the end of the content.

Semantic role: frame. Compiles to eHTML <frame>.

Registers:

• Canonical: <frame>

The simplest case. The handler emits a <figure> wrapper (the vocab html_output.element frame is only the lookup key for handler-strategy entries — the handler controls the actual element). +border is default on for <frame>, so the class appears automatically.

<frame type=note title="Important" |
Make sure to read this carefully.
>

With a title rendered at the top.

<frame #fig:method-box caption="Workflow steps" |
1. Collect data.
2. Clean.
3. Model.
>

Numbered by default (#272). Shares the figure counter with <fig>/<svg>/<mermaid>/<abc>; use -numbered for a frame that is not numbered.

Semantic role: minipage. Compiles to eHTML <minipage>.

Registers:

• Canonical: <minipage>

The simplest case. The handler emits a <figure> wrapper (the vocab html_output.element minipage is only the lookup key for handler-strategy entries — the handler controls the actual element). +border is default on for <minipage>, so the class appears automatically. The body renders as sealed eHTML.

<minipage #mp:compare caption="Side-by-side comparison" |
A figure here counts privately.

<fig #fig:left src="left.png" | Left panel.>
<fig #fig:right src="right.png" | Right panel.>
>

A figure here counts privately.

Numbered by default (#272). It counts in its own “Minipage N” series — <ref @mp:compare> resolves to “minipage 1”. The two inner figures number 1 and 2 in the minipage’s PRIVATE figure counter, NOT the document’s: a document <fig> elsewhere is unaffected, and an outside <ref @fig:left> is a normal not-found ref-error (the seal forbids inbound references to the body).

Semantic role: svg. Compiles to eHTML <svg>.

Registers:

• Canonical: <svg>

<svg -numbered viewBox="0 0 100 100" width=200 height=200 |
  <circle cx="50" cy="50" r="40" fill="blue" />
>

A lone inline SVG. <svg> is numbered by default (it shares the figure counter), so -numbered is what opts out of framing for a purely inline graphic. The source is opaque pipe content; the attributes pass through to the rendered <svg> element.

<svg #fig:diagram viewBox="0 0 100 100" caption="A simple circle" |
  <circle cx="50" cy="50" r="40" />
>

Captioned and numbered → framed by the ordinary frameable rule: the handler wraps the <svg> in a <figure> with the <figcaption> as a sibling inside the wrapper (figcaption is not a valid child of <svg>). Shares the figure counter with <fig>; <ref @fig:diagram> resolves to “Figure N”.

Semantic role: table. Compiles to eHTML <table>.

Registers:

• Canonical: <table>

| Name | Price |
|------|-------|
| foo  | 1     |
| bar  | 2     |

Plain markdown table syntax (via remark-gfm). The most common authoring path for simple tables. No explicit enscribe tags needed.

<csv | name,price
foo,1
bar,2
>

The <csv> shorthand lowers to <table csv> (the standalone <csv> / <tsv> tags were retired to these gate shorthands) and renders today — the table above is its output, parsed from the CSV source by the table handler’s csv engine. <csv | ... > and the qualifying form <table csv | ... > converge to the same parsed table; only the authoring shorthand differs. JSON data uses the qualifying <table json | ... > form — there is no standalone <json> shorthand. See the <csv> vocabulary entry for engine attributes (header control, alignment, etc.).

<table #revenue type=results>
  <caption | Quarterly revenue>
  <tr><th>Quarter</th><th>Revenue</th></tr>
  <tr><td>Q1</td><td>$100M</td></tr>
  <tr><td>Q2</td><td>$120M</td></tr>
</table>

Explicit table with cells, used when fine control over structure or attributes is needed.