Wc

Charts that move with your stack, not against it. One rendering engine, twenty-one chart types, and first-class support for React, Vue, Svelte, Angular, native web components, or no framework at all. Every chart is interactive, accessible, and emits an LLM-ready data context you can drop straight into a report, a dashboard, or an AI feature.

Live demos and docs -> michi-vz.netlify.app

Why @michi-vz

• For developers - pick your framework, keep your charts. The same engine powers React, Vue, Svelte, Angular, web components, and vanilla JS, so a stack change never means a charting rewrite. Fully typed, tree-shakeable, light DOM, and styleable with ordinary CSS.
• For teams and managers - one charting standard across every project and every framework your teams use. No vendor lock-in, MIT-licensed, accessible by default (every chart ships a hidden screen-reader table), and ready for the AI features on your roadmap.
• For analysts, finance, and reporting - purpose-built financial views like tornado (dual horizontal bar), range, gap, and ribbon charts, with a built-in notion of certain vs. estimated values so forecasts and actuals read clearly. Interactive tooltips and highlighting make the numbers easy to explore and easy to put in front of a client.
• For everyone making charts - clean defaults that look good out of the box, smooth interactions, and copy-paste examples in the docs.

Lineage

@michi-vz is the framework-agnostic successor to the original React-only michi-vz (live demos). That first version was React-only; this rebuild extracts the same battle-tested charts into a plain-TS engine so they run on React, Vue, Svelte, Angular, web components, or plain JavaScript, while keeping the original behaviour and visual style intact.

What's inside

A plain-TS rendering engine (@michi-vz/core), native web components (@michi-vz/wc, Lit + light DOM), thin wrappers for React / Vue / Svelte / Angular, and an opt-in client-side AI insights layer (@michi-vz/insights, in progress). Every chart emits a renderer-agnostic, LLM-ready ChartContext (structured data + stats + rule-based NL summary + a hidden a11y table mirror) that is identical whether the chart renders as SVG or canvas.

Charts (21)

GapChart · LineChart · FanChart (forecast) · AreaChart · ScatterPlot · VerticalStackBar · ComparableHorizontalBar · ComparableVerticalBar · DualHorizontalBar (tornado) · BarBell · Range · Ribbon · Radar · Treemap (realized/untapped split + mobile stack) · Pie / Donut · Bubble (gravity cluster + realized/untapped split) · Sankey (flow diagram) · Fountain (Jet d'Eau, experimental) · ChoroplethMap · SymbolMap · RadialTree - each available from the core engine, the web component, and all four framework wrappers, in SVG, canvas, and experimental WebGPU renderers.

Packages

All published to npm under the @michi-vz scope. Every package renders the same charts from the same engine - pick the one that matches your stack:

Install

npm i @michi-vz/core            # the engine (no framework deps)
npm i @michi-vz/wc              # native web components
npm i @michi-vz/react           # or /vue, /svelte, /angular

Usage

Web component (any framework / no build)

<script type="module" src="https://cdn.jsdelivr.net/npm/@michi-vz/wc/dist/michi-vz-wc.bundle.js"></script>
<michi-vz-line-chart id="c" width="600" height="300" chart-title="Sales"></michi-vz-line-chart>
<script>
  const c = document.getElementById("c");
  c.dataSet = [{ label: "North", series: [{ date: 2016, value: 10, certainty: true }, /* … */] }];
  c.addEventListener("michi-vz:highlight", (e) => console.log(e.detail));
  const ctx = c.getContext(); // LLM-ready context, identical in SVG and canvas
</script>

React

import { LineChart } from "@michi-vz/react";

<LineChart dataSet={data} xAxisDataType="date_annual" detectGaps renderer="svg" />;

Imperative engine (no framework)

import { mountLineChart } from "@michi-vz/core";
const chart = mountLineChart(el, { dataSet, width: 600, height: 300 });
chart.update(nextProps);
chart.getContext();
chart.destroy();

Docs for AI assistants

The whole library ships as machine-readable docs (llmstxt.org convention), regenerated on every docs build:

• llms.txt - compact index: packages, guides, all 21 charts
• llms-full.txt - the full reference in one file: per-chart props, usage in every framework, ChartContext, insights, theming, gotchas

Point a coding agent at either URL (also in-repo at apps/docs/public/).

Hard rules (light DOM, colour contract)

Charts render into light DOM only - the canvas renderer resolves mark colours by reading consumer CSS via getComputedStyle on probe elements, so consumer rules like .line[data-label-safe="North"] { stroke: … } reach canvas pixels. Every mark carries data-label + data-label-safe (sanitized via the single sanitizeForClassName). @michi-vz/core/styles.css does layout/tooltip only and never sets fill/stroke.

Develop

pnpm install
pnpm build && pnpm typecheck && pnpm test
pnpm verify:playground   # headless-browser self-tests (expect "N/N checks passed.")

Monorepo: pnpm workspaces + Turborepo + Changesets + tsup. Run apps/docs via Docker.

License and brand

Code: MIT (LICENSE). The michi-vz name, the Michi shield logo, and the cat artwork are not part of the MIT grant - see the brand policy. Redistributed forks must rebrand; fixes and features are welcome upstream via issues and pull requests.