Demo

You can run a local MCP server that exposes a numeric extraction tool and generates chart-ready data from text. This server follows the Model Context Protocol, letting AI agents discover available tools, initialize the session, and call tools to analyze content and visualize results.

How to use

You interact with the MCP server through an MCP client. Start by initializing the connection to obtain the server capabilities, then fetch the list of available tools, and finally call the tool that extracts numbers and summarizes them as chart data. Use the tool to feed in your text content and specify the chart type you want (bar, pie, or line). The server will return the extracted numbers, their frequency, and the data needed to render the requested chart.

How to install

Prerequisites: you need Node.js installed on your machine. A package manager is also required to run the project scripts.

Step-by-step commands you can run to set up and start locally:

Notes and tips

The server exposes an HTTP MCP endpoint for programmatic access and supports standard development and production startup flows. Use the dev flow to iterate quickly, or build and run in production mode for stable deployments.

Available tools

extract_numbers

Checks user input for numbers, extracts them, analyzes their frequencies, and generates chart-ready data.