Matplotlib

This is the foundation of Python visualization. If you need publication-quality figures with precise control over every axis, label, and spine, this is what you reach for. The skill emphasizes the object-oriented interface (fig, ax = plt.subplots()) over the pyplot state machine, which is the right call for anything beyond throwaway plots. It's verbose compared to Seaborn but gives you full customization for scientific papers, multi-panel layouts, heatmaps, and LaTeX-formatted labels. The anti-patterns section is honest about common mistakes like mixing interfaces and using rainbow colormaps. For interactive dashboards or quick statistical plots, look elsewhere, but for static figures that need to be perfect, this is still the standard.

Install to Claude Code

npx -y skills add tondevrel/scientific-agent-skills --skill matplotlib --agent claude-code