Chartsmith

Use cases

Software that has numbers,
and a human about to read them.

Almost every caller here is an agent or a pipeline that already knows the answer and needs to show it. One call turns the rows it is holding into a picture the recipient understands at a glance.

01Picking the form

Let the job pick the chart.

The commonest mistake is choosing a chart type first and bending the data into it. Start from what the numbers have to say.

A magnitude, across a handful of named things
bar — or horizontalBar when the names are long enough to crowd an x axis.
Change over time
line. Use area when the quantity accumulates from zero.
A total and what it is made of
stackedBar.
Parts of one whole, few enough to compare
donut, or pie without a centre total.
Whether two measures move together
scatter.
One number that is the whole story
Not a chart. Put the number in the text — a chart of one value is decoration.
02Real output

Every form, drawn just now.

Rendered when this page loaded, by the same code your API call runs.

Wire it into whatever already has your numbers — REST, MCP, or a keyless x402 call.

Read the docs