> ## Documentation Index
> Fetch the complete documentation index at: https://plasma-ai.mintlify.site/llms.txt
> Use this file to discover all available pages before exploring further.

# Charts and dashboards

> Turn data into readable comparisons, trends, and summaries.

Ask an agent to combine charts with an explanation of what they show. A useful chart identifies its source, units, time period, and any missing data.

## Available building blocks

Pages include time-series charts, bar charts, heatmaps, candlestick charts, and a generic chart component for supported specifications. Smaller components include metrics, sparklines, targets, and status grids.

A dashboard can combine these with tables, filters, sections, and expandable details. Controls depend on how the Page was authored; not every chart includes every interaction.

| Question | Useful view |
| - | - |
| How did a number change over time? | Time-series chart, with a metric for the latest value |
| Which categories differ? | Bar chart and a comparison table |
| Where are concentrations or gaps? | Heatmap |
| How close are we to a goal? | Target and progress metric |
| What are the underlying observations? | Table alongside the chart |

## Example requests

> Create a Page showing weekly signups from this CSV. Include a trend chart, total signups, and a table of the source rows. Label the date range and mark incomplete weeks.

> Compare incident counts by service. Add a bar chart and a heatmap by service and week. Explain whether the data measures counts or rates.

> Build a dashboard with a region filter. Make the chart and table use the same filtered dataset.

For numeric comparisons, specify whether you want totals, averages, percentages, or cumulative values. Ask the agent to state how it handles blanks and duplicates.

## Data and freshness

A chart can use inline values, a saved dataset version, or supported connected data. These are different refresh models.

Publishing a newer saved dataset version does not retarget an existing Page. Ask the author to update its binding when you want a refreshed report. Connected views should identify freshness and show unavailable data clearly.

Chart specifications do not load arbitrary network URLs. For external data, have the agent retain the data or build an authorized [interactive Page](/fractal-product/pages/interactive).

## Explore and review

Use the controls supplied by the author. Filters and sorting usually change the displayed view, not the source data.

Ask for a textual summary and source table when the chart will inform a decision. Use Page comments to flag unclear labels or questionable conclusions; a passage comment anchors to selected text, not automatically to a chart point.

See [Tables and datasets](/fractal-product/pages/data) for versioning and [Evidence and sources](/fractal-product/pages/evidence) for provenance.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.