microsoft/flint-chart ? reverse-engineered prompt

Build me a charting toolkit that helps AI generate nice looking charts from simple chart specs instead of fiddly chart settings.

I want one TypeScript library that can take a compact, human editable chart description plus some data, and turn it into ready to render outputs for Vega Lite, ECharts, Chart.js, Plotly, and Excel charts. It should handle the annoying stuff automatically, like sizing, spacing, labels, legends, and layout, based on the data and the chart type. Make it easy to use with a few examples for common charts like scatter plots, bar charts, line charts, and heatmaps.

Also add an MCP server so an AI assistant can create, validate, and render charts from chat or an editor. It should be able to read pasted rows or local CSV, TSV, or JSON files, and return either a preview image or the native chart spec. If you need to check the latest docs for any chart backend, look them up online.