---
name: statistical-briefing
description: "Use this skill to produce government-style statistical briefings from tabular data: trend summaries, rankings, category comparisons, charts, concise written interpretation, Word briefings, and PowerPoint-ready outputs. Use it when the user asks for 数据简报, 统计分析, 图表解读, 汇报材料, or data-backed briefing slides."
---

# Statistical Briefing

Use this skill after data has been profiled or cleaned. The deliverable should connect tables, charts, and concise administrative interpretation.

## Default Deliverables

- `briefing.docx`: key findings, chart captions, caveats, and recommended follow-up.
- `chart.png`: a stable static chart suitable for Word/PPT.
- Optional `.pptx` if the user asks for slides.

## Script

By default, the script reads `/inputs/manifest.json` and picks the first uploaded `.xlsx`, `.xls`, `.csv`, or `.tsv` file. Pass `--input` only when selecting a specific path from the manifest.

```bash
python /skill/sys_statistical-briefing/scripts/build_briefing.py \
  --category 地区 \
  --value 数量 \
  --out-docx /outputs/statistical_briefing.docx \
  --out-chart /outputs/statistical_chart.png
```

## Interpretation Rules

- State facts before conclusions.
- Avoid causal claims unless the data supports them.
- Include denominator and time range when available.
- Make chart titles specific, not generic.
- Mention data quality caveats that affect interpretation.