Where inline content works

Micronodes are not narrative-only — every mark label is rich text, so formatting and maths reach the canvas.

#Two places inline content is accepted

Micronodes are easy to mistake for a narrative-only vocabulary, because that is where a reader first meets them. They are not. Inline content is accepted in two places: prose, and any attribute whose declared type is rich-text — which in practice means every mark's label:. The same **bold**, $maths$ , !teal{…}, link and !value that work in a paragraph work on the canvas.

canvas.graph{
    curve#bell(expression: exp(-x^2 / 2); label: $N(\mu, \sigma^2)$)
    point(x: 1; y: 2; label: !teal{turning point})
    fit#f(model: polynomial; degree: 1)
    text(x: 0; y: 4; label: slope = !value:#f.slope(digits: 2))
}

#Which attributes are rich text

Eighteen, and every one of them is a label: — on curve, point, scatter, histogram, integral, text, axis.label, the line-like marks (line, segment, vector, polygon) and all seven shapes. Every reference page names the type of each of its attributes, so rich-text in the type column is the reliable signal — this list is what it currently adds up to, not a rule of its own.

#What a label cannot hold

Rich text is inline content, not a body policy. A label renders formatting, maths, colour and live values; it cannot hold a paragraph, a list, a callout or another mark. Where a canvas genuinely needs block content, the scene's detail accepts it directly — paragraph, the four headings, list, equation, table and image are all legal there as well as in the narrative.

This is the practical reason the manual documents one content vocabulary rather than a narrative set and a canvas set: there is only one, and the difference between the two slots is a short list of narrative-only elements — the four callouts, divider, step and the internote card.

Where inline content works

Micronodes are not narrative-only — every mark label is rich text, so formatting and maths reach the canvas.

#Two places inline content is accepted

canvas.graph{
    curve#bell(expression: exp(-x^2 / 2); label: $N(\mu, \sigma^2)$)
    point(x: 1; y: 2; label: !teal{turning point})
    fit#f(model: polynomial; degree: 1)
    text(x: 0; y: 4; label: slope = !value:#f.slope(digits: 2))
}