Article Visual Planner

Article Visual Planner

Use this skill when the user wants a complete visual plan for an article rather than one standalone image. This skill does not define a visual style. It reads article content, decides which visual assets are needed, and chains every asset to a selected cover-* style skill.

Preferred Claude CLI syntax:

/article-visual-planner:cover-cream-orange-knowledge-poster

This means:

• Planner: article-visual-planner
• Style skill: cover-cream-orange-knowledge-poster

Also support field syntax:

使用 $article-visual-planner 自动规划文章配图
文章：{article_path_or_content}
视觉风格：{cover_style_skill}
输出类型：{out_type}
平台：{platform}
资产范围：{asset_scope}

Chain Contract

The selected style must be a cover-* skill. Treat every cover-* skill as a visual style generator, even if its legacy wording says "cover". Pass each planned asset to the selected style with an explicit 资产类型 field:

• cover: article cover, header, hero, or share image.
• article-inline: normal article inline visual.
• workflow-diagram: process, flow, or tutorial step.
• comparison: before/after, old/new, pros/cons, risks/tradeoffs.
• architecture: system stack, module map, data flow, Agent/RAG/tooling diagram.
• long-infographic: dense vertical or poster-like article summary.
• social-card: square or vertical excerpt card.

When outputting prompts directly, apply the selected style skill's visual system to each asset brief and adapt composition to the asset type.

Output Type

Use the explicit --out-type parameter to decide what to output.

• --out-type template: output the invocation template only.
• --out-type brief: output the visual asset plan and chain briefs.
• --out-type prompt: output final image prompts for all planned assets.
• --out-type all: output briefs first, then final prompts.
• Omitted --out-type: default to brief.

If the user asks to generate images directly, produce the plan first unless they explicitly ask to generate one concrete asset now.

Template Output

For template-only mode, output exactly this structure and fill every field:

使用 $article-visual-planner 自动规划文章配图
文章：{article_path_or_content}
视觉风格：{cover_style_skill}
输出类型：{template|brief|prompt|all}
平台：{platform}
资产范围：{asset_scope}
语言：{language}
配图数量：{visual_count}
补充背景：{context}

Required Inputs

Extract or infer:

• Article: path, Markdown content, outline, or topic. Required.
• Style skill: a cover-* skill. Required; if omitted, ask for the style.
• Output type: template, brief, prompt, or all. Default brief.
• Platform: X article, WeChat article, blog, newsletter, documentation, tutorial, or infer.
• Asset scope: default cover + inline visuals.
• Language: Chinese, English, or mixed Chinese-English; infer from article.
• Content visual count: default max(article content images, structure count, 5) unless user gives an exact count.
• Extra context: optional.

Source Article Analysis

When the user provides an article path or Markdown content:

1. Read the article.
2. Count image references: Markdown `, HTML `, or equivalent.
3. Treat the first image as the existing cover when it appears at the top or before the first major heading.
4. Count remaining images as content visuals (N_article).
5. Identify major sections and infer N_structure: sections whose argument benefits from a visual.
6. Use max(N_article, N_structure, 5) as the content visual count unless the user gives an exact count.
7. Map the cover to the full thesis and each content visual to one section-level idea.
8. If a source section already has an image, preserve that editorial judgment by assigning a replacement visual to that section.

Brief Output

For brief mode, output exactly these sections:

【文章分析】
标题：
平台：
语言：
原文图片总数：
封面图片：
内容配图数：
结构推断配图数：
最终配图数：

【链式调用设置】
规划器：$article-visual-planner
视觉风格：${cover_style_skill}
调用格式：/article-visual-planner:{cover_style_skill}

【封面 brief】
调用：${cover_style_skill}
资产类型：cover
主题：
说明：
画幅比例：
用途：
内容结构：
视觉重点：
补充背景：
禁用元素：

【配图 brief 1】
调用：${cover_style_skill}
资产类型：
章节：
主题：
说明：
画幅比例：
用途：
内容结构：
视觉重点：
补充背景：
禁用元素：

【文章配图编排】
配图分布逻辑：
章节到配图映射：
图注语气：
阅读节奏：

【一致性约束】
术语统一：
标题语气：
风格一致：
重复规避：

Repeat the 配图 brief section for the final content visual count.

Prompt Output

For prompt or all mode, convert every planned asset into a provider-neutral image-generation prompt using the selected cover-* style. Do not mention provider names, model names, or runtime-specific syntax.

Every prompt must include:

1. Asset type.
2. Exact canvas ratio and recommended size.
3. Topic, article section, use case, and language.
4. The selected style skill name.
5. Section-specific content structure.
6. Visual focus and text hierarchy.
7. Style-specific visual system adapted from the selected cover-* skill.
8. A concise avoid list.

Use this output shape:

【封面 prompt】

【配图 prompt 1/N】

【文章配图映射】

Defaults

• Cover ratio: X article 5:2, WeChat cover 2.35:1, blog/tutorial 16:9.
• Inline visual ratio: default 16:9.
• WeChat inline ratio: 4:3.
• Social card ratio: 1:1 or 4:5.
• Long infographic ratio: 3:4 or 4:5.

Guardrails

• Do not invent a visual style. Always chain to a selected cover-* style.
• Do not produce fewer content visuals than the article already has.
• Do not make every content visual a mini cover.
• Do not repeat the same full-article map in every section.
• Keep product names, terms, feature names, and stage labels identical across all assets.
• If the user provided /article-visual-planner:{style}, preserve that style name exactly in every chain brief.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: imartinstudio
• Source: imartinstudio/cover-prompt-skills
• License: MIT

Install and usage instructions live in the source repository linked above.