wanghetommy/ichartjs

What it solves

iChart.js is designed to bridge the gap between AI agents and data visualization. It provides a structured, renderer-independent runtime that allows agents to move from raw data to a validated, rendered chart without guessing configuration fields or chart types.

How it works

The library provides a set of public APIs (a "planning contract") that an agent can follow to ensure accuracy. The workflow typically involves:

1. Capability Discovery: The agent calls getCapabilities() to see what the library can actually do.
2. Data Inspection: Using inspectData(), the agent analyzes the dataset.
3. Planning: The agent uses planChart() to determine the best chart type and encoding based on the user's intent.
4. Validation: The agent runs validateSpec() to ensure the generated JSON specification is correct before rendering.
5. Rendering: The agent calls createChart() to render the visualization using either SVG or Canvas.
6. Explanation: The agent uses chart.explain() to describe the result to the user.

Who it’s for

• AI Agent Developers: Those building agents that need to generate professional, validated data visualizations.
• Frontend Developers: Developers integrating a visualization runtime into web applications where AI agents can operate.
• Coding Agents: Agents that can modify project code to implement charts.

Highlights

• Agent-First Design: Includes a machine-readable capability manifest and recipes for AI agents.
• Renderer Independent: Supports both SVG and Canvas rendering.
• Export Options: Capable of exporting to JSON, SVG, PNG, and JPEG.
• Broad Support: Includes foundational charts (Line, Bar, Pie, etc.), project views (Gantt, Timeline), and diagrams (Flow, Swimlane).
• Zero Production Dependencies: The runtime has no external production dependencies.