Microcharts

What ganapativs/microcharts MCP server does

ganapativs/microcharts MCP server gives an MCP client access to the microcharts catalog and rendering workflow. It covers 106 small chart types, including sparklines, bars, bullets, and other chart forms designed for compact interface locations such as messages, tables, KPI cards, and streamed responses.

The server supports a question-first workflow. An agent can describe the analytical intent in plain language, such as checking a trend, showing part-to-whole, or representing an error budget. It can then inspect a matching chart type, obtain the exact inputs, and render the result without building a React component itself.

How it works

The server runs on the local machine and communicates with MCP clients over stdio. Its tools are backed by the microcharts library and return structured information that can be used in subsequent calls.

A typical sequence is:

1. Call find_microchart with a plain-language question about the data. The response ranks candidate chart types and explains why each candidate matched.
2. Call get_microchart with a chart slug. The response contains the chart's import paths, chart-specific and shared props, expected data shape, usage guidance, a runnable example, and a sample object.
3. Pass the sample or adapted values to render_microchart. Supply the series through data and place additional settings such as value, target, curve, color, or width in props.

The final render is a self-contained SVG with styles included in the output. The response also provides generated alt text based on the chart data.

Setup and configuration

Install and run ganapativs/microcharts MCP server with the npm package command below:

npx -y @microcharts/mcp

An MCP client configuration uses npx as the command and -y @microcharts/mcp as its arguments. The provided material describes local stdio execution and does not specify environment variables, API credentials, or hosted deployment settings.

The underlying chart library is available separately as @microcharts/react, but the MCP package is the component described here. The server is intended to let an assistant call the catalog and renderer directly rather than requiring the client to execute React code.

Tools and capabilities

• find_microchart: ranks chart types against an analytical question and includes a reason for each match.
• get_microchart: returns wiring details, props, data expectations, guidance on suitable and unsuitable uses, an example, and JSON-ready sample props.
• render_microchart: creates a finished SVG and generated alt text from a selected chart and its data.

Each chart has its own data shape, so retrieving the sample from get_microchart is the reliable way to prepare a render call. This makes ganapativs/microcharts MCP server a fit for agents that need to select and produce compact visualizations dynamically.

Limitations and notes

The material only documents stdio execution through the npm package. It does not identify specific MCP clients as compatible, describe authentication, or provide configuration for remote hosting. The chart catalog is broad, but not every familiar visualization is included: the project explicitly does not ship pie charts, needle gauges or speedometers, batteries, waffles, or violin charts. When using a chart, follow the returned data shape rather than assuming all chart types accept the same series format.