Intelligrapher
科研绘图智能助手/Intelligent Assistant for Scientific Research Illustration — Claude Code Skill
A Claude Code skill plugin that generates publication-quality academic charts as ready-to-run Python code, encoding domain conventions, journal color palettes, and matplotlib templates so researchers skip manual formatting work.
Intelligrapher is a skill plugin for Claude Code that generates publication-quality research charts as runnable Python code. The problem it addresses is the time researchers spend writing and tuning matplotlib code (the standard Python charting library) every time they need a figure for a paper, especially getting colors, axis labels, fonts, and formatting to match what academic journals expect. The README is written in Chinese and describes the tool as solving this frustration by encoding domain knowledge, aesthetic rules, and code templates into the agent's instructions.
The skill has three internal libraries. The knowledge base contains field-specific conventions for each supported research domain, covering which chart types are standard, how axes should be labeled, and what notation conventions apply (for example, survival curves in medicine must include patient count labels). The aesthetic library stores color palettes extracted from real papers published in top journals, such as deep blue and brick red from Engineering Structures, or the color sets used by The Lancet and NEJM. The template library contains five Python files with placeholder variables for chart type, colors, labels, and output format.
When you describe a chart in plain language, Claude reads the relevant knowledge and aesthetic files, selects the appropriate template, fills in the placeholders, and returns a complete Python script ready to run. You can point it at an existing data file or ask it to generate example data. Supported chart types include line plots with error bands, grouped bar charts with error bars, heatmaps, scatter plots with regression lines, and box plots. Output formats include PNG at specified resolution, SVG, PDF, and TIFF.
Two slash commands are included. The main intelligrapher command handles chart generation. A companion intelligrapher-updater command lets you switch the active domain or add a new one by initializing skeleton files that you fill with conventions and colors relevant to your field. Civil engineering and medicine are included by default.
The skill is installed by cloning the repository into the .claude directory of your project. No external services or API keys are required beyond Claude Code itself.
Where it fits
- Generate a survival curve with patient count labels for a medical paper without writing matplotlib code from scratch.
- Create a grouped bar chart using The Lancet color palette, correctly formatted for a clinical journal figure.
- Add a new research domain by running the updater command and filling in skeleton convention and color files.
- Produce a scatter plot with regression line from a CSV file and export it as a publication-ready SVG or PDF.