Turns data into publication-ready charts using matplotlib, seaborn, or plotly. Give it query results, a CSV, or pasted data and it picks the right chart type (line for trends, bars for comparisons, heatmaps for matrices) or uses what you specify. The workflow handles the full cycle: querying if needed, cleaning data into DataFrames, generating the visualization with proper formatting, and saving as PNG. What's nice is the opinionated design layer. It applies actual best practices like starting bar chart axes at zero, formatting large numbers as "1.2M" instead of "1200000", and sorting categories by value instead of alphabetically. The code it produces is readable Python you can modify yourself.
npx -y skills add anthropics/knowledge-work-plugins --skill create-viz --agent claude-codeInstalls into .claude/skills of the current project.
Select a file.
juliusbrussee/caveman
mattpocock/skills
shadcn/improve
obra/superpowers
forrestchang/andrej-karpathy-skills
vercel-labs/skills