README

This repository includes a reusable agent skill for publication-ready scientific plots, with a small demo project.

Structure

• scientific-plotting/ is the skill package. Import it to your workspace as needed.

• demo/ is a small demo project for trying the skill quickly:

  1. Open a terminal in demo/

  2. Ask the AI to plot data from data/

  3. The plotting script is written to scripts/

  4. The figure is exported to output/

Usage

Example prompt (Flexible, free-form natural language):

Read data from the `data/` folder and create a scientific plot using only raw data.
Set the x-axis range to `1e1` - `1e8`, use the skill color palette, and export to `output/`.

Example Results

• Both Chinese and English prompts work.
• Model: Codex GPT-5.4, medium reasoning
• Results generated by running in the demo/ folder

Prompt 1:

读取 `data` 文件夹中的数据，仅使用 raw 数据进行科研绘图。将横坐标范围设为 1e1 - 1e8，使用 skill 中的配色，并将结果导出到 `output/`。

Prompt 2:

Read data from the `data/` folder and use only raw data for scientific plotting.
Set the x-axis range to `1e1` - `1e8`, use the color scheme defined by the skill, and export the results to `output/`.

With the scientific-plotting skill, the LLM maintains a consistent plot style across two independent trials.

The data was acquired directly from a electrical spectrum analyzer using a delayed self-heterodyne setup with a 5 µs delay, characterizing both one RIO PLANEX™ 1550 nm and a TOPTICA CTL 1550 lasers. Please note that the measurement conditions do not necessarily reflect the optimal linewidth performance of these lasers.

Limitations

• Current support is limited to basic curve plots.
• Image size constraints are currently insufficient; the plotting area dimensions may not be uniform across exported figures.
• Advanced plot types, including heatmaps, violin plots, and 3D visualizations, are currently unavailable.