Chart choice, consistent styling and journal-ready export for matplotlib, seaborn and plotly.
Runcell teamIllustrative output. The real run happens in your notebook.
Width: 3.5 in (single column) Fonts: 8 pt minimum, embedded Palette: colorblind-safe (Okabe-Ito) Export: figure2.pdf (vector) + figure2.png (300 dpi) Caption draft: written
Agents can draw a chart in one call. The hard part is drawing the right chart, styled consistently, at the size and resolution a journal, a slide deck or a dashboard actually needs. This skill carries those rules so the agent does not reinvent them each time.
It includes a publication mode: figure widths matched to single or double column, font sizes that survive shrinking, colorblind-safe palettes, vector export, and a caption checklist. In Jupyter the figure is regenerated from the cell, so it always matches the current data.
Recommended by
Runcell team
The idea came from a manuscript that was returned for figure sizing, twice. Publication mode encodes the boring rules, column widths, minimum font sizes, vector export, so the figure is right the first time. Because it runs in a notebook, the figure is regenerated from the cell whenever the data changes.
It gives the agent rules for making charts well: which chart fits the data, how to style it consistently, and how to export it at the right size and resolution. Claude Code can call matplotlib without a skill, but the results vary from run to run. The skill makes them consistent.
That is the point of publication mode. Journals specify column widths, minimum font size, resolution and file type. The skill encodes common defaults, for example 3.5 inches for a single column, 7 inches for double, 300 dpi, and vector output where accepted, and you can set a specific journal's numbers. Always check the final submission guidelines yourself.
matplotlib and seaborn by default, plotly when you need interactivity. The style rules are library-independent, so the same skill works whichever one you choose.
Yes. It defaults to palettes that remain distinguishable under common color vision deficiencies and avoids red-green contrasts for categorical data.
A systematic first pass over any dataset: structure, quality, distributions, relationships, and what to do next.
Runcell teamFreeLiterature search, citation verification and structured notes for agents that write with references.
Runcell teamFreeRead, clean and analyze spreadsheets with pandas. Reproducible answers, not a rewritten sheet.
Runcell teamFreeRuncell is a Jupyter-native AI agent that executes every step in a live kernel. Results are cells you can inspect and rerun.