Use to make an article 图文并茂 — generate publish-ready figures (data charts, flow/structure diagrams) and a cover, SEO/GEO-friendly. 配图/图表/插图/封面/流程图/数据图/visuals/figures/diagrams/charts
Scanned 8/30/2026
Install to Claude Code
npx -y skills add weitzu-com/ai-article-factory --skill article-visuals --agent claude-codeInstalls into .claude/skills of the current project.
Are you the author of Article Visuals?
Add the live security badge to your README — it updates automatically with every re-scan.
[](https://www.skillsdirectory.com/skills/weitzu-com-article-visuals)More formats (shields.io, HTML) on the badges page.
---
name: article-visuals
description: 'Use to make an article 图文并茂 — generate publish-ready figures (data charts, flow/structure diagrams) and a cover, SEO/GEO-friendly. 配图/图表/插图/封面/流程图/数据图/visuals/figures/diagrams/charts'
version: "1.0.0"
license: Apache-2.0
homepage: "https://github.com/weitzu-com/ai-article-factory"
when_to_use: "Use at the figures stage (P6) of article production, or whenever a draft needs charts/diagrams/a cover. Renders Mermaid + matplotlib to PNG and prepares an AI-cover prompt; enforces image SEO/GEO rules."
argument-hint: "<article-dir>"
metadata:
author: weitzu-com
tags: [visuals, figures, charts, diagrams, mermaid, matplotlib, image-seo, cover]
---
# article-visuals (P6 — 图文并茂)
Make figures that serve three readers at once: humans (understanding), search engines (image SEO / rich results), and AI engines (multimodal citation). **≥3 figures per article: 1 cover + ≥2 content figures.** This skill is self-contained — it uses the bundled scripts, no external pack needed.
## Type decision tree (pick type → tool)
| To express | Type | Tool | Source file |
|---|---|---|---|
| flow / structure / relationship | diagram | **Mermaid** | `assets/src/*.mmd` |
| data / comparison / trend | chart | **matplotlib** (`import figlib`) | `assets/src/*.py` |
| concept / cover | illustration | **AI image** (prompt → human fills) | `assets/src/cover-prompt.md` → `assets/img/cover.png` |
| real UI / object | screenshot/photo | manual | — |
| tabular comparison | **Markdown table** (preferred — AI-extractable) | in prose | — |
**Rule:** if data fits a Markdown table, use a table; only draw when relationships/trends need it.
## Workflow
1. From the outline's figure intent, write figure sources into `assets/src/` and log `assets/图清单.md` (caption + alt + placement).
2. Render:
```bash
bash <repo>/scripts/render-figures.sh <article-dir> # mermaid + matplotlib → assets/img/
```
`figlib.py` provides the brand palette (`scripts/palette.json`) + registered CJK fonts (Chinese won't box) + `save()` / `bar_compare()` helpers. If `mmdc` is missing, `.mmd` is skipped gracefully.
3. Embed with `` + a caption carrying the source `[C#]`.
4. Cover: a placeholder is used so it never blocks; generate the real one from `cover-prompt.md` and drop it at `assets/img/cover.png`.
## Image SEO + GEO checklist (each figure must pass)
- Descriptive English kebab-case filename with keywords (not `image1.png`).
- Alt text describing the figure + naturally containing the keyword.
- Caption is a complete sentence with the data source.
- Compressed (<200KB chart / <400KB cover).
- **Every data figure has an equivalent sentence/table nearby** (AI reads text; the figure corroborates).
- Don't rely on color alone (color-blind safe); in-figure font ≥ body.
Full detail: `../../references/visuals-spec.md` (the repo-root `references/` folder).
Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.
No comments yet. Be the first to comment!