The usual path for this is a model writing pandas code that you then paste into a notebook, which puts two manual steps between a question and an answer. Loading the file and rendering the chart inside the server removes both. The documentation is thin, so expect to learn the shape of the analysis by asking rather than by reading — the file formats and the two plotting libraries are what it commits to.
A data analysis server for local spreadsheet files. It loads `.xlsx` and `.csv`, runs statistical analysis on what it finds, and renders charts through matplotlib and plotly — so a question about a file is answered with a figure rather than a code snippet you then have to run.
- Loading and analysing `.xlsx` and `.csv` files
- Statistical analysis over the loaded data
- Charts rendered with matplotlib
- Interactive charts rendered with plotly
- Local operation over files already on the machine, with nothing uploaded anywhere
Python with `uv`. Clone the repository and run `uv sync`, then have the client launch it with `uv --directory` pointed at the checkout, or `uvx mcp-data-analyzer` for the published package — version 0.1.8. Transport is stdio.
One command — uvx mcp-data-analyzer