This is an early release preview. You may encounter bugs.
Flint Chart logo
Unclaimed

Tool interface

Flint Chart

πŸͺ„ Flint is a visualization language that lets AI agents reliably create expressive, good-looking charts from simple, human-editable chart s

Built by Microsoft

B 74/100 GitHub score ? This grade is derived from GitHub signals, not user votes. Open for the full breakdown.
Discussion

01 / Summary

About Flint Chart.

Flint is a visualization intermediate language from Microsoft Research for agents that generate charts. Instead of having an agent hardcode scales, axes, spacing, labels, and layout, Flint takes a compact semantic chart spec, the data, and an optional visual theme, derives those decisions itself, and compiles the result to a native Vega-Lite, ECharts, Chart.js, or Plotly spec, or to an editable Excel chart.

The spec records what each field means through more than 70 semantic types (such as rank, temperature, price, or country), and layout adapts sizing, labels, marks, and legends to the data's cardinality and the canvas. A theme spec sits beside the chart spec and defines layout behavior, semantic presentation, and visual identity once for a whole chart library. Ten built-in presets include editorial styles modeled on the New York Times, the Economist, Swiss design, and McKinsey, and a custom theme can inherit a preset and override only its brand decisions. Themes currently apply to Vega-Lite and Plotly output.

The project has two parts. The flint-chart JavaScript and TypeScript library takes the same input for every backend and returns that library's native spec object. The flint-chart-mcp server, local or hosted, lets agents choose a chart template, validate it, and open an interactive chart view, static PNG or SVG output, or backend-native specs from a chat or coding environment, with data passed inline or read from local JSON, CSV, or TSV files. A standalone agent skill covers agent workflows without MCP, and a Python port is in preview.

  • Semantic chart specs: more than 70 semantic types describe what each field means
  • Automatic layout: sizing, spacing, labels, marks, and legends adapt to data cardinality, chart design, and canvas size
  • Visual themes: ten presets plus custom themes that inherit a preset and override selected decisions
  • Multiple backends: one input compiles to Vega-Lite, ECharts, Chart.js, Plotly, or native editable Excel charts
  • MCP server: agents create, validate, and render charts, with interactive views, PNG or SVG output, or native specs
  • Hosted MCP endpoint: works with remote-MCP clients such as GitHub Copilot in VS Code and Claude
  • Local data input: rows passed inline or read from JSON, CSV, or TSV files
  • Agent skill: a standalone chart-authoring skill for agents that do not use MCP
  • Dynamic chart widgets: switch chart types and edit chart properties in place
  • Live editor and gallery: a project site with examples, a theme explorer, and an editor

02 / Discussion CREDIBILITY-GATED

Discussion

Reading is open to everyone. Posting and voting need a verified identity or a GitHub grade of B or higher.

  • No discussions yet.

04 / Build

Build with Flint Chart.

Browse the catalog for frameworks, tools, and harnesses, each scored on real GitHub credibility.

Get Flint Chart β†’

Browse the catalog