Flint Interactive:AI 代理人以語義規格快速建立互動圖表
Creating interactive data visualization with Flint-interactive
Flint Interactive 讓 AI 代理人只需簡短的語義規格,即可產出多種互動圖表,支援 52 種圖表型別、5 個渲染後端及 14 預設主題。
AI 伴隨問答
針對這篇報導有疑問?可直接點選快捷問題或向 AI 追問背景脈絡與技術細節:
Flint is a visualization intermediate language that lets AI agents reliably create expressive, good-looking and interactive charts from simple, human-editable specs. The key design is to disentangle the visual, style and interaction logic that low-level chart libraries mix together: you or your agents describe each in a short semantic spec, and the compiler composes them.
Flint supports 52 chart types across 5 rendering backends, 14 themes and 23 interaction presets. For bespoke visualizations, design your own themes in the Theme Lab and build advanced interactions with the Flint API. Use it as an MCP server with your agent, or as an npm library for building interactive applications.
Charts
Interactions
Applications
Flint spec
{
"data": {...},
"semantic_types": {
"period": "YearMonth",
"newUsers": "Profit"
},
"chart_spec": {
"chartType": "Waterfall Chart",
"encodings": {
"x": "period",
"y": "newUsers"
},
"title": "How the player base moved through 2025",
"subtitle": "Net new players each month, accumulating into the year's total"
},
"theme_spec": "bbc",
"interaction_spec": {
"interactions": [{"type": "click-highlight"}]
}
}
Theme:
Corners 0
Values
Try it now
Copy this prompt into your AI agent, such as GitHub Copilot, Claude Code, Cursor or Codex. It adds the Flint MCP server and draws your first chart.
Wondering about the experience? Check out the MCP server.
Help me set up Flint as an MCP server named "flint"!
1. Add it to my client config, using whichever option my client supports:
- a local (stdio) server: npx -y flint-chart-mcp
- the hosted server over HTTP: https://flint.data-formulator.ai/mcp
2. Verify the setup by listing the available Flint chart types.
3. Show me a fun first chart: how many moons each planet has, in a Flint theme you like.
Updates
October 19, 2026
Flint 1.0 introduces Flint Interactive: interaction presets, semantic events, and an interaction API that connects charts to your app.
VideoInteractiveAugust 12, 2026
Flint 0.5.1 extends visual themes to the Plotly backend and adds Vega-Lite Calendar Heatmap.
ChangelogAugust 5, 2026
Flint 0.5.0 introduces formal visual themes with ten presets, custom ThemeSpec authoring, and preset inheritance.
How it works?
A Flint spec consists of the data, semantic types, the chart spec, and optional theme and interaction specs. Instead of relying on complex user-side logic to glue them together, the compiler does the heavy lifting: it resolves each part against the others and composes a complete backend-native spec (shown here in Vega-Lite), filling in the necessary low-level details, and renders a good-looking, interactive chart.
Flint spec
{
"data": {...},
"semantic_types": {
"Month": "Month",
"City": "City",
"Temperature": "Temperature"
},
"chart_spec": {
"chartType": "Heatmap",
"encodings": {
"x": "Month",
"y": "City",
"color": "Temperature"
}
},
"theme_spec": "nyt",
"interaction_spec": {
"interactions": [{
"type": "hover-group-focus",
"options": {
"groupBy": "City"
}
}]
}
}
Compiled spec (Vega-Lite)
{
"data": {...},
"mark": {
"type": "rect",
"stroke": "#ffffff",
"strokeWidth": 1
},
"width": {
"step": 28.23333333333333
},
"height": {
"step": 28.23333333333333
},
"encoding": {
"x": {
"field": "Month",
"type": "ordinal",
"sort": ["Jan", "Feb", "Mar",
"Apr", "May", "Jun",
"Jul", "Aug", "Sep",
"Oct", "Nov", "Dec"],
"axis": {"title": null}
},
"y": {
"field": "City",
"type": "nominal",
"sort": null,
"axis": {"title": null}
},
"color": {
"field": "Temperature",
"type": "quantitative",
"scale": {
"domain": [-30, 30],
"range": ["#2f6b9a",
"#8fb4cc",
"#efece5",
... // 66 more lines (excluding interaction)
Visualization
Flint spec
{
"data": {...},
"semantic_types": {
"Month": "Month",
"City": "City",
"Temperature": "Temperature"
},
"chart_spec": {
"chartType": "Heatmap",
"encodings": {
"x": "Month",
"y": "City",
"color": "Temperature"
}
},
"theme_spec": "nyt",
"interaction_spec": {
"interactions": [{
"type": "hover-group-focus",
"options": {
"groupBy": "City"
}
}]
}
}
Compiled spec (Vega-Lite)
{
"data": {...},
"mark": {
"type": "rect",
"stroke": "#ffffff",
"strokeWidth": 1
},
"width": {
"step": 28.23333333333333
},
"height": {
"step": 28.23333333333333
},
"encoding": {
"x": {
"field": "Month",
"type": "ordinal",
"sort": ["Jan", "Feb", "Mar",
"Apr", "May", "Jun",
"Jul", "Aug", "Sep",
"Oct", "Nov", "Dec"],
"axis": {"title": null}
},
... // 80 more lines (excluding interaction)
Visualization
1.Describe data with semantic types
Data fields in Flint are described with semantic types (e.g., Rank, YearMonth, Delta, Temperature) and their intrinsic information, such as domains and units. The Flint compiler uses them to infer parsing, scales, axes, formatting, and colors.
Compiled chart
2.Build charts with library-agnostic chart specs
Charts are defined with high-level chart types and encodings. The compiler automatically optimizes the low-level settings based on the data's characteristics, for example as the data gets denser. The same spec renders across Vega-Lite, ECharts, Chart.js, and Plotly, or as native Excel charts.
Dense · 22 × 3
3.Unify the look with a theme spec
Geometry, typography, and color are defined as reusable formal specs that apply across the whole library, independent of the data and chart type. Pick one of 14 preset themes, or write your own for coherent branding.
Swiss
4.Add interactions with an interaction spec
Flint provides 23 interaction presets and a programming API for building interactive charts into applications. The compiler automatically resolves user interaction events into semantic information, so your app only needs to answer with high-level chart updates.
Live · try it
Start building with Flint.
Open source and ready to use. Start from GitHub or browse examples in the gallery.
Join the Flint WeChat group
Scan the QR code to join the Flint user WeChat group and chat with other users.

Join the Flint Discord
Ask questions, share your charts, and follow releases with Flint users everywhere.