Agent Skills
› KunAgent/Kun
› data-visualization
data-visualization
GitHub用于设计和生成准确的可读统计可视化图表，包含规范的比例尺、标签和注释。通过工具路由渲染图表或运行代码，确保数据验证、图表选择及结果检查的准确性与诚实性。
触发场景
需要生成统计图表
数据可视化需求
安装
npx skills add KunAgent/Kun --skill data-visualization -g -y
SKILL.md
Frontmatter
{
"name": "data-visualization",
"description": "Design and generate accurate statistical visualizations with readable scales, labels, annotations, and source context."
}
Data Visualization
Copyright (c) 2026 KunAgent. Licensed under the MIT License.
Purpose
Design and generate accurate statistical visualizations with readable scales, labels, annotations, and source context.
Tool routing
| Tool or skill | Use |
|---|---|
render_chart |
Render governed charts directly from structured data. |
bash |
Run Python visualization code for file-based figures. |
design_create_diagram |
Create chart-like explanatory diagrams on canvas. |
Workflow
- Identify the analytical question and comparison structure.
- Validate data types, units, missing values, and aggregation.
- Select the simplest chart that preserves the relationship.
- Use direct labels, honest scales, accessible color, and source notes.
- Export at the requested dimensions and inspect the result.
Completion gates
- Check totals, ordering, axes, units, legends, and annotations against source data.
- Use zero baselines where magnitude comparison requires them.
- Avoid dual axes unless the relationship and scale are unambiguous.
Boundaries
- Do not smooth, truncate, or aggregate in ways that change the conclusion without disclosure.
- Do not use decorative charts when a table or metric is clearer.
Delivery
Lead with the outcome, name the evidence used for verification, and disclose any real limitation that remains.
版本历史
- a984912 当前 2026-09-09 01:49
