Skip to content
TopicTracker
From HackerNewsView original
TranslationTranslation

Show HN: Better Graphs – Teach agents to stop making plain Matplotlib slop

Better Graphs is an open-source project that teaches good Matplotlib visualization practices to both AI agents and humans. It includes agent instructions, a tutorial blog, and opinionated default styles (minerva.mplstyle), aiming to produce cleaner, more professional plots inspired by Tufte's design principles.

Background

- **Matplotlib** is the most popular data-visualization library in Python, but its default settings produce ugly, cluttered, publication-unready charts — often called "Matplotlib slop" by practitioners. - **Edward Tufte** is a legendary statistician and information designer whose book *The Visual Display of Quantitative Information* (1983) pioneered minimalist, data-dense charting with principles like avoiding "chartjunk" (unnecessary ink, 3D effects, excessive gridlines). - **data-to-viz.com** and **python-graph-gallery.com** are free web resources cataloguing chart types and best practices; **matplotlib-journey.com** is a paid premium tutorial site known for stunning Matplotlib output. - The project provides reusable **Claude Code skills** (a format for instructing Anthropic's Claude AI coding agent) and a **CLAUDE.md** file (a project-level config that tells the AI how to behave in that codebase) so that LLM-powered coding agents automatically produce good-looking, Tufte-inspired charts instead of raw defaults.