How to use the Matplotlib Plotting agent skill

What does the Matplotlib skill make the agent do?

Once installed, the agent follows this skill's guidance when building charts in Matplotlib. It knows how to build line, scatter, bar, histogram, heatmap, contour, box and 3D plots, arrange multi-panel figures, customise colours and fonts, and save image files in PNG, PDF or SVG.

The skill targets Matplotlib 3.11.2 and recommends the object-oriented interface (Figure and Axes) for most work. It also ships Python scripts and style presets the agent can run to generate example plots or custom style sheets.

• Plot types: line, scatter, bar, histogram, heatmap, contour, box, violin and 3D
• Multi-panel layouts using subplots, mosaic layouts or GridSpec
• Colour, font and style customisation, including rcParams
• Export to PNG, PDF and SVG for publication

How do you install the Matplotlib skill?

Download the skill as a ZIP file from this page and unzip it. You can install it for yourself or for a single project:

For all your projects, unzip it into ~/.claude/skills/. For one project only, unzip it into the .claude/skills/ folder inside that project.

In the Claude apps, you can instead upload the ZIP under Settings → Capabilities → Skills.

• Personal install: ~/.claude/skills/
• Project install: .claude/skills/ inside your project folder
• Claude apps: Settings → Capabilities → Skills, then upload the ZIP

What do you need before you start?

The skill requires Python 3.11 or newer and Matplotlib 3.11.2. The bundled examples also use NumPy and SciPy. pandas is only needed for the pandas examples, and ipympl is only needed for Jupyter widgets. The source suggests uv for installation, and installing packages needs network access.

For project work, the source shows this setup command:

• Python 3.11+ and Matplotlib 3.11.2 (required)
• NumPy and SciPy (for bundled examples)
• pandas (optional, for pandas examples)
• ipympl (optional, for Jupyter widgets)
• uv (suggested for installation)
• No API keys or accounts are needed for local plotting

How do you trigger the skill with a natural request?

You do not need special syntax. Describe the chart you want in plain language, and mention your data and output format. The agent uses the skill when the request involves creating or customising a plot, building a multi-panel figure, exporting visuals, or working with 3D or interactive plots.

Be specific about the file type and quality you need. For quick statistical charts or interactive dashboards, the skill itself points you to seaborn or plotly instead.

• "Plot these measurements as a scatter plot with a fitted line and save it as a 300 dpi PDF."
• "Make a 2x2 figure of my experiment results with shared axis labels and export it as SVG."
• "Convert this old Matplotlib script to the 3.11 API, especially the box plot arguments."

What examples and scripts are included?

The skill includes reference guides for the API, plot types, styling and common issues, plus two scripts: a plot template and a style configurator. The plot template provides executable synthetic examples. The style configurator is for generating custom style sheets.

The guide notes that fragment examples assume imports and named data, so adapt and validate them before use.

• references/api_reference.md: core classes and methods
• references/plot_types.md: examples for each plot type
• references/styling_guide.md: styling and colormap guidance
• references/common_issues.md: display and backend troubleshooting
• scripts/plot_template.py: runnable example plots
• scripts/style_configurator.py: custom style sheet generation

What safety points should you check?

This skill runs scripts and uses the network. The agent will execute Python files on your computer, and installing packages requires network access. Review the bundled scripts before letting the agent run them, especially on machines with sensitive data.

File output needs no GUI. For batch scripts, the source recommends setting the backend to Agg so no window is opened. Interactive windows need a GUI toolkit such as PySide6 or Tk, and these depend on your environment.

• Read scripts/plot_template.py and scripts/style_configurator.py before the first run
• Use MPLBACKEND=Agg for batch scripts that only save files
• GUI windows, Jupyter widgets and external LaTeX were not tested by the source's reviewers and depend on your environment

What tips help you get better figures?

Ask for the object-oriented style and name your figure size and export settings. When migrating older code, tell the agent to follow the 3.11 conventions: use tick_labels and orientation for box plots, mpl.colormaps[name] for colormaps, and label contour lines rather than using contourf.

Save at 300 dpi with bbox_inches='tight' for print, and use vector formats such as PDF or SVG when the figure will be scaled.

• Use tick_labels and orientation for box plots in 3.11
• Use mpl.colormaps[name] to look up colormaps
• Use bbox_inches='tight' to avoid clipped labels
• For journal-styled multi-panel figures, use the scientific-visualization skill instead