# anyviz

Official site: https://anyviz.aiware.store/
Repository: https://github.com/TseringYuu/anyviz
npm: https://www.npmjs.com/package/anyviz
License: MIT
Current version: 1.0.1

## What anyviz is

anyviz is an AI-native data visualization specification and workflow library. It helps AI coding assistants and developers produce professional charts by turning visualization judgment into reusable rules: chart selection, aesthetic parameters, rendering adapters, cross-chart consistency, and accessibility checks.

anyviz is not just one charting runtime. It is a visualization grammar that can be applied across Web, Python, and R output targets.

## Dual visual modes

The anyviz homepage demonstrates two primary visual worlds that use the same underlying grammar:

- Dashboard / Chartscape: dark, immersive, real-time, high-density visualization for monitoring screens, operational dashboards, city maps, finance views, IoT systems, and executive canvases.
- Academic / Editorial: light, precise, publication-grade visualization for papers, reports, analytical documents, and Tufte-inspired charts.

The dual-mode design communicates the core promise: one visualization grammar can produce many professional visual worlds.

## Core workflow

anyviz uses a five-stage workflow:

1. Analyze: infer data shape, analytical intent, audience, and context.
2. Aesthetics: apply theme parameters for color, typography, spacing, grid lines, strokes, labels, and data-ink ratio.
3. Adapt: choose the rendering engine based on the environment and chart type.
4. Consistency: keep colors, labels, units, axes, legends, typography, and formatting coherent across multiple charts.
5. Accessibility: check contrast, color-blind safety, redundant encodings, readable labels, and non-color-only meaning.

## Template library

anyviz includes 34 production-oriented chart templates across these families:

- Statistical charts: line, area, bar, stacked bar, scatter, heatmap, donut, box plot, distribution, and related analytical charts.
- Maps: choropleth maps, flow maps, bubble maps, and geospatial views.
- Graphs and hierarchy: network diagrams, Sankey charts, treemaps, sunburst charts, and relationship views.
- 3D: globe, surface, and 3D scatter templates.

Use templates by describing the data fields, semantic meaning, unit, intended question, audience, and output environment.

## Theme system

anyviz provides four preset themes:

- modern: clean web and product interfaces.
- analytics: business reports, presentations, and analytical dashboards.
- dashboard: dark real-time monitoring and command-center displays.
- academic: paper-friendly, print-oriented, restrained figures.

Themes define palettes, text hierarchy, grid behavior, line width, bar radius, chart background, and contrast behavior. Unspecified chart properties inherit from the selected theme to preserve consistency.

## Rendering adapters

anyviz can guide output for seven rendering targets:

- D3.js: default Web adapter for custom SVG and precise control.
- ECharts: Web dashboards and fast interactive production charts.
- Mapbox: geospatial and high-precision map visualization.
- Three.js: 3D scatter, surface, globe, and immersive scenes.
- Plotly: interactive Python and notebook workflows.
- Matplotlib: static Python charts for publication and export.
- ggplot2: R statistical visualization.

## Natural-language customization

Natural-language instructions are mapped to controlled aesthetic parameters. For example:

- "Make it suitable for a realtime dashboard" maps to dashboard theme, dark background, stronger glow, and monitoring-style hierarchy.
- "Make it look like a paper figure" maps to academic theme, grayscale-safe palette, reduced grid, and serif-friendly typography.
- "Warm colors but keep consistency" adjusts accent palette while inheriting the global typography and spacing rules.

The goal is to modify intent while preserving the rest of the visualization system.

## Accessibility and consistency

anyviz prefers:

- direct labels over unnecessary legends when readable;
- color-blind-safe palettes;
- redundant encodings such as shape, line style, annotation, or symbols;
- consistent color identity for the same entity across charts;
- readable contrast and typography;
- restrained grid lines and high data-ink ratio.

## Quick start for AI assistants

When using anyviz with an AI assistant, provide:

- the dataset or schema;
- the analytical question;
- the target audience and medium;
- desired theme if known;
- target runtime such as D3, ECharts, Plotly, Matplotlib, or ggplot2;
- whether multiple charts must share a canvas or dashboard.

Example prompt:

"Visualize this sales dataset with anyviz. Use the dashboard theme, choose the right chart types, output ECharts code, explain the chart-selection rationale, and keep colors, labels, units, and typography consistent across all charts."

## Resources

- Homepage: https://anyviz.aiware.store/
- LLM context: https://anyviz.aiware.store/llms.txt
- GitHub repository: https://github.com/TseringYuu/anyviz
- npm package: https://www.npmjs.com/package/anyviz
- Skill entrypoint: https://github.com/TseringYuu/anyviz/blob/main/SKILL.md
- Issues: https://github.com/TseringYuu/anyviz/issues