infographic-powerpoint-deck

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

Infographic PowerPoint Deck

信息图风格PowerPoint演示文稿

Default input model (raw article first)

默认输入模式（优先原始文章）

This skill should work directly from the user's raw article content by default.

If the user provides a full article, sermon, report, or long notes:

Read the source content and infer the audience, burden, density, and argument/story structure.
Build
```
slides_message_plan.md
```
from the article structure.

Turn

slides_message_plan.md

into

slides_display_plan.md

Turn

slides_display_plan.md

into

slides_visual_plan.md

Write one image prompt per slide and generate the deck.

If the user already provides

slides_message_plan.md

or an equivalent content-first slide table, you can skip the article-intake step and use that directly.

If the user provides one or more reference slides/screenshots, use

references/reference_slide_intake.md

to extract the reusable visual grammar before writing prompts.

本skill默认支持直接基于用户提供的原始文章内容工作。

如果用户提供完整的文章、讲稿、报告或长篇笔记：

读取源内容，推断受众、核心诉求、信息密度、论点/故事结构。
基于文章结构生成
```
slides_message_plan.md
```
。

将

slides_message_plan.md

转化为

slides_display_plan.md

。

将

slides_display_plan.md

转化为

slides_visual_plan.md

。

为每页幻灯片编写图片提示词，生成演示文稿。

如果用户已经提供

slides_message_plan.md

或等效的内容优先幻灯片表格，可以跳过文章摄入步骤，直接使用该材料。

如果用户提供一张或多张参考幻灯片/截图，在编写提示词前请使用

references/reference_slide_intake.md

提取可复用的视觉规范。

Artifact model

工件模型

Use three distinct planning artifacts:

```
slides_message_plan.md
```
- content / presenter planning only
- what each slide is meant to teach, explain, or say in full
- the detailed speaking content or narrative burden behind the slide
- source/article anchors, scripture/evidence anchors, and must-preserve claims
- what beats must stay separate
```
slides_display_plan.md
```
- what will actually be shown on the slide
- title, verse fragment, module headers, bullets, labels, captions, footer references
- explicit on-slide text after text-budget and archetype decisions
- can be much shorter than the message plan
```
slides_visual_plan.md
```
- downstream visual-production planning
- deck archetype
- style pack choice
- layout routing
- scene choice
- text budget and other production constraints

slides_message_plan.md

is the source of truth for what the presenter means to communicate.

slides_display_plan.md

is the source of truth for what text the audience actually sees.

slides_visual_plan.md

is the source of truth for design and prompt production.

使用三类独立的规划工件：

```
slides_message_plan.md
```
- 仅用于内容/演讲者规划
- 每张幻灯片需要传达、解释的完整内容
- 幻灯片背后的详细演讲内容或叙事诉求
- 源/文章锚点、引用/证据锚点、必须保留的核心观点
- 必须拆分展示的内容节点
```
slides_display_plan.md
```
- 幻灯片上实际展示的内容
- 标题、引用片段、模块标题、项目符号、标签、说明文字、页脚参考
- 经过文本容量和原型决策后的明确展示文字
- 内容长度通常远短于内容规划
```
slides_visual_plan.md
```
- 下游视觉制作规划
- 演示文稿原型
- 风格包选择
- 排版路由
- 场景选择
- 文本容量和其他制作约束

slides_message_plan.md

是演讲者想要传达内容的唯一可信来源。

slides_display_plan.md

是观众实际看到的文字内容的唯一可信来源。

slides_visual_plan.md

是设计和提示词制作的唯一可信来源。

Quick start (default: images-only deck, structured planning)

快速开始（默认：纯图片演示文稿、结构化规划）

Start from either:
- raw article content / notes and use
```
references/article_intake_workflow.md
```
  , or
- a prebuilt
```
slides_message_plan.md
```
  , or
- an underspecified content slide table that you will normalize first
- optional reference slide images / screenshots and use
```
references/reference_slide_intake.md
```
If starting from a raw article, create
```
slides_message_plan.md
```
first.
Infer one deck archetype from the approved message plan using
```
references/deck_archetype_routing.md
```
.
- If the user explicitly requests a style, use that as an override.
Pick one style pack folder.
- If user does not specify style and archetype does not strongly imply a different choice, default to
```
editorial-light
```
  .
- Style pack chooses the deck's visual language (palette, lighting, typography attitude, scene bias). It does not force one fixed layout for every slide.
Create
```
slides_display_plan.md
```
from
```
slides_message_plan.md
```
.
- Decide what the audience should actually see on each slide after considering archetype, density, and likely layout family.
- Do not force the full speaking content onto the slide.
Create
```
slides_visual_plan.md
```
from
```
slides_display_plan.md
```
.
- Add style, layout, scene, and text-budget decisions here, not in the message plan.
Gather the style-pack material:
- You may run
```
scripts/compose_style_pack_blocks.py --pack-id <id>
```
  as a helper, or manually read the style-pack files.
- Treat the script output as helper context only, not as the final prompt body.
- The concrete prompt should still be authored by the model in plain instruction prose.
For each slide in
```
slides_visual_plan.md
```
, fill
```
references/prompt_template.md
```
:
- Use the style-pack material as helper context and rewrite it into plain instruction prose.
- Write the prompt instructions in English by default.
- Paste exact required on-slide text from
```
slides_display_plan.md
```
  in the user-specified language(s).
- Add slide-specific scene layer details and icons using the visual plan.
- Keep internal routing IDs such as
```
L1
```
  or
```
L10
```
  in planning metadata only. Translate the routed layout into plain composition language inside the concrete image prompt.
- If using a reference image or edit-style tool call, keep that mechanic in the tool call. The concrete prompt should still read as result-oriented art direction such as
```
Using the provided image as the compositional base...
```
  , not tool language such as
```
edit this image
```
  .
Generate each slide as one 16:9 image.
- Allowed slide-creation modes in this skill are image generation and image editing/reference only.
- In the slide prompt text, explicitly include a ratio lock sentence (e.g.,
```
Hard canvas constraint: 16:9 widescreen. Do not generate a square image.
```
  ).
- In the tool call, explicitly set ratio config when supported (e.g.,
```
generation_config
```
  with
```
aspect_ratio: "16:9"
```
  ).
- If consistency needs reinforcement, you may use an approved prior slide image or background exploration image as an input/edit reference, but the image tool must still render the final full slide image.
QA readability + text accuracy; regenerate only failed slides.

If any slide is not 16:9 or has text defects, regenerate the image directly.
Do not crop, pad, resize, or post-process generated images unless the user explicitly asks.
Do not render text afterward with Python/PIL/PPT overlays as part of the normal skill workflow.

Assemble images-only PPTX with
```
scripts/build_images_only_pptx.py
```
.

Outputs to produce in the user’s workspace:

```
slides_message_plan.md
```
(detailed per-slide speaking / teaching plan; source of truth for what each slide is meant to communicate)
```
slides_display_plan.md
```
(audience-facing on-slide text plan; source of truth for displayed text and labels)
```
slides_visual_plan.md
```
(visual-production plan; source of truth for archetype/style/layout/scene decisions)
```
prompts.md
```
(the exact prompts used for each slide)
```
slides/slide01.png ...
```
(final 16:9 images)
```
deck_images_only.pptx
```
(concatenated images)

基于以下任意一种输入开始：
- 原始文章内容/笔记，使用
```
references/article_intake_workflow.md
```
  ，或
- 预先制作的
```
slides_message_plan.md
```
  ，或
- 你需要先标准化的未明确内容的幻灯片表格
- 可选的参考幻灯片图片/截图，使用
```
references/reference_slide_intake.md
```
如果基于原始文章开始，首先创建
```
slides_message_plan.md
```
。
使用
```
references/deck_archetype_routing.md
```
从已确认的内容规划中推断出演示文稿原型。
- 如果用户明确要求某种风格，优先使用该风格。
选择一个风格包文件夹。
- 如果用户没有指定风格，且原型没有明确指向其他选择，默认使用
```
editorial-light
```
  。
- 风格包决定演示文稿的视觉语言（调色板、光影、排版风格、场景倾向），不会强制所有幻灯片使用固定排版。
基于
```
slides_message_plan.md
```
创建
```
slides_display_plan.md
```
。
- 结合原型、信息密度和可能的排版类型，决定观众在每张幻灯片上实际需要看到的内容。
- 不要把完整的演讲内容都堆在幻灯片上。
基于
```
slides_display_plan.md
```
创建
```
slides_visual_plan.md
```
。
- 将风格、排版、场景、文本容量决策放在这里，不要放在内容规划里。
收集风格包材料：
- 你可以运行
```
scripts/compose_style_pack_blocks.py --pack-id <id>
```
  作为辅助工具，或手动读取风格包文件。
- 脚本输出仅作为辅助上下文，不要作为最终提示词主体。
- 具体提示词仍需要由模型以纯指令文本的形式编写。
针对
```
slides_visual_plan.md
```
中的每张幻灯片，填充
```
references/prompt_template.md
```
：
- 将风格包材料作为辅助上下文，改写为纯指令文本。
- 默认用英文编写提示词指令。
- 从
```
slides_display_plan.md
```
  中复制用户指定语言的精确展示文字。
- 使用视觉规划添加幻灯片专属的场景层级细节和图标。
- 仅在规划元数据中保留内部路由ID（如
```
L1
```
  或
```
L10
```
  ），在具体图片提示词中将路由排版转换为普通的排版描述语言。
- 如果使用参考图片或编辑风格工具调用，将该逻辑放在工具调用部分。具体提示词仍应使用面向结果的美术指导表述，例如
```
使用提供的图片作为排版基础...
```
  ，而不是工具类表述如
```
编辑这张图片
```
  。
将每张幻灯片生成为一张16:9图片。
- 本skill允许的幻灯片创建模式仅为图片生成和图片编辑/参考。
- 在幻灯片提示词文本中明确加入比例锁定语句，例如
```
硬画布约束：16:9宽屏。不要生成方形图片。
```
  。
- 在工具调用中，如果支持的话明确设置比例配置，例如包含
```
aspect_ratio: "16:9"
```
  的
```
generation_config
```
  。
- 如果需要加强一致性，可以使用已确认的历史幻灯片图片或背景探索图片作为输入/编辑参考，但图片工具仍需渲染最终的完整幻灯片图片。
质检可读性和文字准确性，仅重新生成不合格的幻灯片。
- 如果任意幻灯片不是16:9比例或存在文字缺陷，直接重新生成图片。
- 除非用户明确要求，否则不要裁剪、填充、调整大小或对生成的图片做后处理。
- 常规skill工作流中不要使用Python/PIL/PPT叠加层后续渲染文字。
使用
```
scripts/build_images_only_pptx.py
```
组装纯图片PPTX。

需要在用户工作区生成的输出：

```
slides_message_plan.md
```
（详细的逐页演讲/教学规划，是每张幻灯片需要传达内容的可信来源）
```
slides_display_plan.md
```
（面向观众的幻灯片文字规划，是展示文字和标签的可信来源）
```
slides_visual_plan.md
```
（视觉制作规划，是原型/风格/排版/场景决策的可信来源）
```
prompts.md
```
（每张幻灯片使用的精确提示词）
```
slides/slide01.png ...
```
（最终16:9图片）
```
deck_images_only.pptx
```
（拼接后的图片演示文稿）

Optional structured slide table input

可选结构化幻灯片表格输入

For higher-quality multi-step workflows, you may start from either a raw article or an already-structured message slide table:

Raw article content is a valid first-class input to this skill. Use
```
references/article_intake_workflow.md
```
.
If the upstream artifact is already
```
slides_message_plan.md
```
, preserve it as the message source of truth.
If the upstream artifact already decides deck length, section coverage, or which beats must stay on separate slides, treat that as the controlling message plan unless the user explicitly asks for further compression or expansion.
If the upstream content table is simple or underspecified, normalize it into
```
slides_message_plan.md
```
using
```
references/slide_table_normalization.md
```
before display planning.

对于更高质量的多步工作流，你可以从原始文章或已经结构化的内容幻灯片表格开始：

原始文章内容是本skill的合法一级输入，使用
```
references/article_intake_workflow.md
```
。
如果上游工件已经是
```
slides_message_plan.md
```
，保留其作为内容可信来源。
如果上游工件已经确定了演示文稿长度、章节覆盖范围或必须拆分到不同幻灯片的内容节点，将其作为核心内容规划，除非用户明确要求进一步压缩或扩展。
如果上游内容表格比较简单或不明确，在做展示规划前使用
```
references/slide_table_normalization.md
```
将其标准化为
```
slides_message_plan.md
```
。

Prompt stack (minimal vs full)

提示词栈（极简版 vs 完整版）

Minimal stack (recommended for most decks):

```
references/article_intake_workflow.md
```
(when starting from raw article content)
```
references/deck_archetype_routing.md
```
(infer the overall deck form first)
```
references/style-pack-catalog.md
```
(choose pack ID)
```
references/style-pack-system.md
```
(load order + composition rules)
```
references/layout_routing_policy.md
```
(auto-route layouts by default)
```
references/prompt_example_library.md
```
(use examples and anti-examples to avoid panel drift)

references/high_fidelity_prompt_playbook.md

(when the user wants a more precise prompt)

```
references/reference_slide_intake.md
```
(when user wants the deck to learn from a reference slide)
```
references/style-packs/<pack-id>/
```
(pack-local blocks)
```
references/typography_spacing_lock.md
```
```
references/text_fidelity_block.md
```
```
references/negative_prompt_block.md
```

Full stack (use when you want maximum narrative + visual control):

Minimal stack +
```
references/storyboard_library.md
```
```
references/shot_mood_library.md
```
```
references/scene-catalog.md
```
```
references/scene_library.md
```
```
references/layout_library.md
```
```
references/chinese_quote_compression.md
```
(when long Chinese quotes need splitting)

极简栈（推荐大多数演示文稿使用）：

```
references/article_intake_workflow.md
```
（从原始文章内容开始时使用）
```
references/deck_archetype_routing.md
```
（首先推断整体演示文稿形式）
```
references/style-pack-catalog.md
```
（选择包ID）
```
references/style-pack-system.md
```
（加载顺序 + 组合规则）
```
references/layout_routing_policy.md
```
（默认自动路由排版）
```
references/prompt_example_library.md
```
（使用正面和反面示例避免排版偏差）
```
references/high_fidelity_prompt_playbook.md
```
（用户需要更精确提示词时使用）
```
references/reference_slide_intake.md
```
（用户希望演示文稿参考现有幻灯片风格时使用）
```
references/style-packs/<pack-id>/
```
（风格包本地模块）
```
references/typography_spacing_lock.md
```
```
references/text_fidelity_block.md
```
```
references/negative_prompt_block.md
```

完整栈（需要最大化叙事和视觉控制时使用）：

极简栈 +
```
references/storyboard_library.md
```
```
references/shot_mood_library.md
```
```
references/scene-catalog.md
```
```
references/scene_library.md
```
```
references/layout_library.md
```
```
references/chinese_quote_compression.md
```
（需要拆分长中文引用时使用）

Style-pack system (modular and scalable)

风格包系统（模块化可扩展）

Use folderized style packs so each style is self-contained:

Each style pack lives in
```
references/style-packs/<pack-id>/
```
.
Each pack has its own
```
manifest.toml
```
and block files that define the look.
Packs can inherit from
```
base-core
```
to reuse shared constraints.
Prompts are authored from: base-core guidance + chosen pack guidance + slide-specific content.
This avoids cross-file mismatch and makes new style creation deterministic.

Note: despite the Bible-themed examples, this workflow works for any topic. Swap scenes/props in

references/scene_library.md

to match your domain (product, research, training, etc.).

使用文件夹化的风格包，每个风格都自包含：

每个风格包存放在
```
references/style-packs/<pack-id>/
```
路径下。
每个包有独立的
```
manifest.toml
```
和模块文件定义视觉风格。
风格包可以继承自
```
base-core
```
复用通用约束。
提示词基于以下内容编写：base-core指引 + 选定风格包指引 + 幻灯片专属内容。
这避免了跨文件不匹配，让新风格创建过程可预期。

注意：尽管有圣经主题的示例，本工作流适用于任何主题。可以替换

references/scene_library.md

中的场景/道具匹配你的领域（产品、研究、培训等）。

Decision model (keep style, layout, and scene separate)

决策模型（风格、排版、场景分离）

Message plan = content-level communication model: what the presenter means to say on each slide and how many slides are needed to say it well.
Display plan = audience-facing text model: what short text, labels, verse fragments, bullets, captions, and module headers should actually appear on the slide.
Deck archetype = deck-level viewing model: what kind of visual presentation best fits the approved message plan.
Style pack = deck-level visual language: palette, lighting, texture, typography attitude, scene bias.
Layout = slide-level composition grammar: split-panel, framework, comparison, warning, didactic teaching board, or full-bleed overlay. It should normally be auto-routed during visual planning. The best result may feel highly organized without looking like a rigid template.
Scene ID / visual scene = slide-level depiction choice chosen during visual planning.
Text budget = production constraint describing how much copy the chosen layout needs to carry.

Use the deck archetype to drive the rest of the decisions:

message plan -> slide roles, source anchors, and coverage depth
message plan + archetype + text budget -> display plan
display plan -> on-slide text and label density
deck archetype -> style pack
deck archetype -> visual-plan layout bias
slide roles + text budget -> layout routing
slide role + source meaning -> scene choice
optional reference slide -> prompt wording, layout bias, and possible style-pack refinement

Slide count is not a deck-skill default. It should be determined by how many slides are actually needed to deliver the content well.

Use this decision rule:

identify the distinct beats that need separate treatment
decide how much compression each beat can tolerate without hurting clarity, persuasion, or usability
set slide count from that coverage need

Upstream slide plans, agent instructions, runtime expectations, or user requests can constrain the deck, but they are not the core planning logic. If an explicit brevity request would force harmful compression, preserve the content logic first and either:

keep the richer deck, or
consciously produce a compressed version while making that tradeoff explicit in
```
slides_message_plan.md
```

Do not collapse a rich article into a short deck merely because the skill can make a shorter deck.

Infographic does not mean one universal text density. Some decks should feel like sparse keynote slides. Other decks should feel like self-contained didactic infographics or teaching posters with more visible words, labels, formulas, diagram captions, and explanatory sections. Choose the density and layout family that best serves the content instead of forcing every deck toward the same low-text aesthetic.

Use one style pack across the deck by default, and let the skill choose layout per slide unless the user explicitly overrides it:

Short emotional or opener slides often route to
```
L4
```
,
```
L5
```
, or
```
L6
```
.
In cinematic, editorial, warm, airy, animated, and youth decks, short/medium context or application slides should try
```
L4
```
or
```
L6
```
before
```
L1
```
if the scene can provide a calm text-safe zone.
Teaching, framework, comparison, and didactic poster slides often route to
```
L1
```
,
```
L2
```
,
```
L3
```
,
```
L7
```
,
```
L8
```
,
```
L9
```
,
```
L10
```
, or
```
L11
```
when structure or density requires those layouts.
The same style pack can legitimately use both split-panel and full-bleed layouts inside one deck.
If you need article-first intake guidance, read
```
references/article_intake_workflow.md
```
.
If you need archetype routing guidance, read
```
references/deck_archetype_routing.md
```
.
If you need routing guidance, read
```
references/layout_routing_policy.md
```
.

内容规划 = 内容层级沟通模型：演讲者在每张幻灯片想要表达的内容，以及需要多少张幻灯片才能清晰传达。
展示规划 = 面向观众的文字模型：幻灯片上实际需要展示的短文本、标签、引用片段、项目符号、说明文字、模块标题。
演示文稿原型 = 演示文稿层级浏览模型：哪种视觉呈现形式最匹配已确认的内容规划。
风格包 = 演示文稿层级视觉语言：调色板、光影、纹理、排版风格、场景倾向。
排版 = 幻灯片层级构图规范：分屏、框架、对比、警示、教学板、全屏覆盖。通常在视觉规划阶段自动路由。最佳效果可能看起来高度有条理，但不会像僵硬的模板。
场景ID/视觉场景 = 视觉规划阶段选择的幻灯片层级呈现方案。
文本容量 = 描述选定排版可承载的文本量的生产约束。

使用演示文稿原型驱动后续决策：

内容规划 -> 幻灯片角色、源锚点、覆盖深度
内容规划 + 原型 + 文本容量 -> 展示规划
展示规划 -> 幻灯片文字和标签密度
演示文稿原型 -> 风格包
演示文稿原型 -> 视觉规划排版倾向
幻灯片角色 + 文本容量 -> 排版路由
幻灯片角色 + 源含义 -> 场景选择
可选参考幻灯片 -> 提示词措辞、排版倾向、可能的风格包优化

幻灯片数量不是本skill的默认配置，应该由实际清晰传达内容需要的幻灯片数量决定。

使用以下决策规则：

识别需要单独处理的不同内容节点
决定每个内容节点在不损害清晰度、说服力、可用性的前提下可接受的压缩程度
根据覆盖需求确定幻灯片数量

上游幻灯片规划、agent指令、运行时预期或用户请求可以约束演示文稿，但它们不是核心规划逻辑。如果明确的简洁要求会导致损害内容的过度压缩，优先保留内容逻辑，可以选择以下任一方案：

保留内容更丰富的演示文稿，或
有意识地生成压缩版本，同时在
```
slides_message_plan.md
```
中明确说明该权衡。

不要仅仅因为skill可以生成更短的演示文稿，就把内容丰富的文章压缩成短PPT。

信息图不代表统一的文本密度。有些演示文稿应该是稀疏的 keynote 风格幻灯片。其他演示文稿可能是自成体系的教学信息图或教学海报，包含更多可见文字、标签、公式、图表说明和解释模块。选择最适配内容的密度和排版类型，不要强制所有演示文稿都使用低文本的审美风格。

默认情况下整个演示文稿使用一个风格包，除非用户明确覆盖，否则skill会为每张幻灯片单独选择排版：

短的情感类或开场幻灯片通常路由到
```
L4
```
、
```
L5
```
或
```
L6
```
。
在电影感、编辑风、温暖、轻盈、动画、青年风格的演示文稿中，短/中等内容或应用类幻灯片如果场景可以提供安静的文本安全区，应该优先尝试
```
L4
```
或
```
L6
```
，再考虑
```
L1
```
。
教学、框架、对比、教学海报类幻灯片在结构或密度需要时，通常路由到
```
L1
```
、
```
L2
```
、
```
L3
```
、
```
L7
```
、
```
L8
```
、
```
L9
```
、
```
L10
```
或
```
L11
```
。
同一个风格包在一个演示文稿中可以同时使用分屏和全屏排版。
如果你需要原始文章摄入指引，阅读
```
references/article_intake_workflow.md
```
。
如果你需要原型路由指引，阅读
```
references/deck_archetype_routing.md
```
。
如果你需要路由指引，阅读
```
references/layout_routing_policy.md
```
。

Slide prompt recipe (copy/paste template)

幻灯片提示词模板（可复制粘贴）

Read

references/prompt_template.md

and fill it per slide. Keep it extremely explicit:

Start from the inferred article structure, message plan, display plan, and deck archetype, not from arbitrary slide decoration.
Put all required on-slide text from
```
slides_display_plan.md
```
into one exact-text block in the concrete prompt.
If the slide is a dense didactic infographic, include every visible section header, label, callout, formula label, diagram caption, and module heading in that exact-text block rather than expecting the model to invent them correctly.
For visuals, describe scene layers (far/mid/foreground), plus 3–8 concrete objects.
For high-fidelity results, do not stop at naming the style pack or the layout family. Also specify geometry, palette/material treatment, typography attitude, divider/module structure, and lighting behavior in plain prose.
Specify what must be subtle/low-contrast so text stays readable.
Keep prompt instructions in English by default even when the required on-slide text is Chinese, German, or bilingual.
Do not clutter concrete prompts with redundant language metadata in the default English case. Only state prompt/output language explicitly when the slide uses non-English or multilingual rendered text, or when the language setup might otherwise be ambiguous.
Use both presentation families intentionally: split-panel / board-style infographic layouts for dense teaching content, didactic poster layouts for self-contained educational slides, and full-bleed cinematic overlay layouts for title cards, quote slides, and scene-led emotional beats.
Treat split-panel wording as explicit layout instructions, not as generic presentation language. If the user prefers text directly on the image, route away from
```
L1
```
before writing the prompt.
Treat
```
L1
```
-
```
L11
```
as internal routing handles, not model-facing prompt text. The image model should receive composition instructions such as
```
use a mirrored two-zone teaching board with a central divider
```
, not
```
use layout L9
```
.
Treat style-pack files the same way: they are internal guidance material, not literal prompt fragments. The final prompt should read like direct art-direction and display instructions, not like serialized metadata.
Treat input-image and edit mechanics the same way: they are tool-side mechanics, not prompt-side prose. The final prompt should describe the desired finished slide, not the fact that an edit tool is being used.
If the final prompt still sounds generic after removing the metadata, enrich it rather than shortening it. High-fidelity prompts normally need explicit visual specificity, not just raw content and one style label.
For full-bleed layouts, default to direct text on the image itself with composition-based readability support. Do not ask for visible rounded rectangles, frosted cards, or caption boxes unless the user explicitly wants that treatment.
Do not ask the user for per-slide layouts by default. Route layouts internally from slide role, text budget, display density, and overall deck style. Only use explicit
```
Layout hint
```
when the user or
```
slides_visual_plan.md
```
provides one.
Do not ask the user for style-pack IDs by default when the article content already makes the fit obvious. Infer archetype first, then select style.

If you need article-first intake guidance, read

references/article_intake_workflow.md

. If you need deck archetype selection guidance, read

references/deck_archetype_routing.md

. If you need inspiration for more cinematic scene material and prop ideas, read

references/scene_library.md

. If you need stable scene IDs + tags for selection/filtering, read

references/scene-catalog.md

. If you need copy-ready visual scene blocks, read

references/scene-preset-library.md

. If you need fast, reliable infographic compositions or full-bleed title-card layouts, read

references/layout_library.md

. If you need automatic routing rules for when to use each layout, read

references/layout_routing_policy.md

. If you need concrete good/bad prompt examples before writing prompts, read

references/prompt_example_library.md

. If you need a higher-precision checklist for turning style/layout/scene choices into a concrete prompt, read

references/high_fidelity_prompt_playbook.md

. If the user gives a screenshot or example slide and wants the skill to learn from it, read

references/reference_slide_intake.md

. If you need shot-language / time-of-day / weather / lighting presets, read

references/shot_mood_library.md

. If you need style modularity, read

references/style-pack-system.md

and use

scripts/compose_style_pack_blocks.py

. If you want fewer typos and more readability, paste these into every slide prompt:

references/typography_spacing_lock.md

references/text_fidelity_block.md

references/negative_prompt_block.md

. If you want storyboard-like viewing rhythm, read

references/storyboard_library.md

. If a slide uses language-specific rendering constraints or long non-Latin quote blocks, read the relevant reference note before writing the concrete prompt (for example,

references/chinese_quote_compression.md

阅读

references/prompt_template.md

并为每张幻灯片填充内容，保持高度明确：

基于推断的文章结构、内容规划、展示规划和演示文稿原型，不要做随意的幻灯片装饰。
将
```
slides_display_plan.md
```
中所有要求的幻灯片文字放在具体提示词的精确文本块中。
如果幻灯片是高密度的教学信息图，在精确文本块中包含所有可见的章节标题、标签、标注、公式标签、图表说明和模块标题，不要指望模型能正确生成这些内容。
对于视觉效果，描述场景层级（远/中/前景），加上3-8个具体对象。
要获得高保真效果，不要只提到风格包或排版类型，还要用普通文本明确说明几何结构、调色板/材质处理、排版风格、分隔线/模块结构、光影效果。
明确说明哪些元素需要微妙/低对比度，保证文字可读性。
默认用英文编写提示词指令，即使要求的幻灯片文字是中文、德文或双语。
默认英文场景下不要在提示词中加入冗余的语言元数据。仅当幻灯片使用非英文或多语言渲染文字，或语言配置可能存在歧义时，才明确说明提示词/输出语言。
有意识地使用两种呈现类型：分屏/板型信息图排版用于高密度教学内容，教学海报排版用于自成体系的教育幻灯片，全屏电影覆盖排版用于标题页、引用幻灯片、场景主导的情感节点。
把分屏措辞作为明确的排版指令，不要作为通用的呈现语言。如果用户更喜欢文字直接放在图片上，在编写提示词前不要选择
```
L1
```
排版。
把
```
L1
```
-
```
L11
```
作为内部路由标识，不要作为面向模型的提示词文本。图片模型应该接收类似
```
使用带中央分隔线的镜像双区域教学板
```
的构图指令，而不是
```
使用L9排版
```
。
风格包文件同理：它们是内部指引材料，不是字面的提示词片段。最终提示词应该读起来像直接的美术指导和展示指令，而不是序列化的元数据。
输入图片和编辑逻辑同理：它们是工具侧的逻辑，不是提示词侧的文本。最终提示词应该描述期望的最终幻灯片效果，而不是说明正在使用编辑工具。
如果移除元数据后最终提示词仍然看起来很通用，要丰富内容而不是缩短。高保真提示词通常需要明确的视觉细节，而不仅仅是原始内容和一个风格标签。
对于全屏排版，默认直接在图片上放置文字，配合基于构图的可读性支持。除非用户明确要求，否则不要添加可见的圆角矩形、磨砂卡片、说明框。
默认不要向用户询问每张幻灯片的排版，根据幻灯片角色、文本容量、展示密度和整体演示文稿风格内部路由排版。仅当用户或
```
slides_visual_plan.md
```
提供明确说明时，才使用
```
排版提示
```
。
当文章内容已经明确适配的风格时，默认不要向用户询问风格包ID。首先推断原型，再选择风格。

如果你需要原始文章摄入指引，阅读

references/article_intake_workflow.md

。如果你需要演示文稿原型选择指引，阅读

references/deck_archetype_routing.md

。如果你需要更多电影感场景素材和道具创意，阅读

references/scene_library.md

。如果你需要稳定的场景ID+标签用于选择/过滤，阅读

references/scene-catalog.md

。如果你需要可直接使用的视觉场景模块，阅读

references/scene-preset-library.md

。如果你需要快速可靠的信息图构图或全屏标题页排版，阅读

references/layout_library.md

。如果你需要自动路由规则决定使用哪种排版，阅读

references/layout_routing_policy.md

。如果你在编写提示词前需要具体的好坏提示词示例，阅读

references/prompt_example_library.md

。如果你需要更高精度的检查清单，将风格/排版/场景选择转化为具体提示词，阅读

references/high_fidelity_prompt_playbook.md

。如果用户提供截图或示例幻灯片，希望skill学习该风格，阅读

references/reference_slide_intake.md

。如果你需要镜头语言/时间/天气/光影预设，阅读

references/shot_mood_library.md

。如果你需要风格模块化能力，阅读

references/style-pack-system.md

并使用

scripts/compose_style_pack_blocks.py

。如果你希望减少错别字提升可读性，将以下内容粘贴到每张幻灯片提示词中：

references/typography_spacing_lock.md

、

references/text_fidelity_block.md

、

references/negative_prompt_block.md

。如果你想要分镜般的浏览节奏，阅读

references/storyboard_library.md

。如果幻灯片有特定语言的渲染约束或长非拉丁引用块，在编写具体提示词前阅读相关参考说明（例如

references/chinese_quote_compression.md

）。

Quality bar (and how to iterate)

质量标准（以及迭代方法）

For each slide:

Message fit first: the deck should reflect the structure and intent of the article and the richer message plan, not just produce isolated pretty slides.
Display discipline second: the slide should show only what the audience needs to see, not the entire speaking content.
Readability first: text never overlaps busy imagery; use either a dedicated text panel or a deliberately clean overlay-safe region, depending on the selected layout.
Auto-layout first: unless the user explicitly overrides, let the skill choose the layout from slide role, text budget, and style rather than forcing one composition family.
No hallucinated text: forbid extra words/logos/watermarks.
Text fidelity in the requested language/script: if any character, accent, word, or punctuation is wrong, regenerate that slide with:
- shorter quote blocks,
- ```
All required on-slide text must be exact. Do not rewrite. Do not add or remove punctuation or spaces.
```
  ,
- simpler backgrounds.
Engagement: add background scenes + textures + icon clusters, but keep them low-contrast.
Didactic density is allowed when the content needs it: a self-contained teaching slide may legitimately carry more visible words than a keynote slide, as long as hierarchy, spacing, and diagram structure stay clear.

Common failure fixes:

Deck feels visually polished but structurally wrong: revisit
```
references/article_intake_workflow.md
```
and the message/display boundary; fix the slide roles or display compression before rewriting prompts.
Tool call fails/intermittent errors: retry after a short wait; keep prompt stable; reduce scene complexity only if repeats.
Text too small: reduce bullet count; split into two slides; demand large type and comfortable line spacing.
Slide feels too empty for an infographic teaching use case: reroute to
```
L9
```
,
```
L10
```
, or
```
L11
```
and explicitly include the needed labels, captions, formulas, or comparison headings instead of pretending the content is sparse.
Too busy: force a low-contrast background, keep the scene mostly on the right, and keep the left panel clean.
Deck too dark: switch to
```
editorial-light
```
or
```
airy-relaxed
```
, disable vignette, force daylight scene IDs, and add brightness override.
Full-bleed overlay feels boxed in: remove visible cards, rounded rectangles, frosted panels, and boxed banners; ask for direct text on the image with only subtle local contrast support.
Panel drift appears on a cinematic/editorial/warm slide: compare against
```
references/prompt_example_library.md
```
, reroute
```
L1
```
to
```
L4
```
or
```
L6
```
, and remove any
```
text panel
```
wording before retrying.
Auto-routed layout feels wrong: reroute the slide using
```
references/layout_routing_policy.md
```
; do not keep forcing a weak layout just because an earlier guess picked it.
Language/script fidelity drifts: shorten copy first, simplify the background, use the relevant language-specific rendering reference, and only then adjust style/layout.
Ratio mismatch: regenerate that slide with stricter 16:9 constraint in prompt; do not crop or otherwise post-process.
- Ensure both prompt text and tool-call config specify 16:9 in the same retry.

针对每张幻灯片：

内容匹配优先：演示文稿应该反映文章和丰富内容规划的结构和意图，而不仅仅是生成独立的好看幻灯片。
展示规范其次：幻灯片应该只展示观众需要看到的内容，不要包含完整的演讲内容。
可读性优先：文字永远不要和复杂背景重叠；根据选定的排版，要么使用专用的文本面板，要么使用特意留出的干净覆盖安全区域。
自动排版优先：除非用户明确覆盖，让skill根据幻灯片角色、文本容量和风格选择排版，不要强制使用单一构图类型。
禁止虚构文字：不允许额外的文字/Logo/水印。
请求语言/脚本的文字保真：如果任何字符、重音、单词、标点错误，重新生成该幻灯片，加入：
- 更短的引用块，
- ```
所有要求的幻灯片文字必须精确。不要改写。不要添加或移除标点或空格。
```
  ，
- 更简单的背景。
吸引力：添加背景场景+纹理+图标簇，但保持低对比度。
内容需要时允许高密度教学内容：自成体系的教学幻灯片可以合理地比keynote幻灯片包含更多可见文字，只要层级、间距和图表结构保持清晰。

常见问题修复：

演示文稿视觉精美但结构错误：重新查看
```
references/article_intake_workflow.md
```
和内容/展示边界；在重写提示词前修复幻灯片角色或展示压缩规则。
工具调用失败/间歇性错误：短暂等待后重试；保持提示词稳定；仅当多次失败时降低场景复杂度。
文字太小：减少项目符号数量；拆分为两张幻灯片；要求大字体和舒适的行间距。
信息图教学场景下幻灯片太空：重新路由到
```
L9
```
、
```
L10
```
或
```
L11
```
，明确加入需要的标签、说明、公式或对比标题，不要假装内容很稀疏。
太杂乱：强制使用低对比度背景，把场景主要放在右侧，保持左侧面板干净。
演示文稿太暗：切换到
```
editorial-light
```
或
```
airy-relaxed
```
，关闭暗角，强制使用日间场景ID，添加亮度覆盖。
全屏覆盖感觉有边界：移除可见卡片、圆角矩形、磨砂面板、边框横幅；要求文字直接放在图片上，仅使用微妙的局部对比度支持。
电影感/编辑风/温暖风格幻灯片出现排版偏差：和
```
references/prompt_example_library.md
```
对比，将
```
L1
```
重新路由到
```
L4
```
或
```
L6
```
，重试前移除所有
```
文本面板
```
措辞。
自动路由排版感觉不合适：使用
```
references/layout_routing_policy.md
```
重新路由幻灯片；不要因为之前的选择就强制使用不合适的排版。
语言/脚本保真偏差：首先缩短文本，简化背景，使用相关的特定语言渲染参考，之后再调整风格/排版。
比例不匹配：在提示词中加入更严格的16:9约束重新生成幻灯片；不要裁剪或做其他后处理。
- 重试时确保提示词文本和工具调用配置都指定了16:9比例。

Operating mode (be patient; avoid “probe images”)

运行模式（耐心操作，避免“探测图片”）

Image generation can be slow. Prefer generating the real slides directly.
Avoid generating test/probe images (e.g. blank backgrounds) unless the user explicitly asks for diagnostics.
If an image call fails intermittently, retry the same slide prompt after a brief pause; do not change multiple variables at once.
Default delivery mode is raw outputs: keep generated images as-is and assemble the PPT directly from them.
If you create a background exploration image to stabilize later generations, treat it only as an image-tool reference input, not as a canvas for later text overlay or any non-image-tool composition step.

图片生成可能很慢，优先直接生成真实幻灯片。
除非用户明确要求诊断，避免生成测试/探测图片（例如空白背景）。
如果图片调用间歇性失败，短暂等待后用相同的幻灯片提示词重试；不要同时修改多个变量。
默认交付模式是原始输出：保持生成的图片原样，直接用它们组装PPT。
如果你创建了背景探索图片来稳定后续生成，仅将其作为图片工具的参考输入，不要作为后续文字叠加或任何非图片工具构图步骤的画布。

Tooling constraints (important)

工具约束（重要）

Some image tools only allow writing output images to specific directories (often

~/Downloads

or a workspace path). When blocked:

Generate images into an allowed directory (e.g.
```
~/Downloads/<deck_id>/
```
).
Copy images into the project/workspace output folder.

有些图片工具仅允许将输出图片写入特定目录（通常是

~/Downloads

或工作区路径）。被限制时：

将图片生成到允许的目录（例如
```
~/Downloads/<deck_id>/
```
）。
将图片复制到项目/工作区输出文件夹。

Bundled scripts

捆绑脚本

scripts/compose_style_pack_blocks.py

scripts/compose_style_pack_blocks.py

Compose inherited style-pack blocks into one paste-ready bundle.
Use before prompt writing to prevent style mismatch.

Run:

bash

python3 scripts/compose_style_pack_blocks.py --pack-id editorial-light

将继承的风格包模块组合为一个可直接粘贴的包。
在编写提示词前使用，避免风格不匹配。

运行：

bash

python3 scripts/compose_style_pack_blocks.py --pack-id editorial-light

scripts/create_style_pack.py

scripts/create_style_pack.py

Scaffold a new style pack folder with manifest + block templates.
Use when existing packs are not enough and you need fast extension.

Run:

bash

python3 scripts/create_style_pack.py --pack-id warm-minimal --display-name "Warm Minimal" --keywords "warm,balanced" --scene-tags "warm,clean"

生成新的风格包文件夹框架，包含清单和模块模板。
现有风格包不足以满足需求，需要快速扩展时使用。

运行：

bash

python3 scripts/create_style_pack.py --pack-id warm-minimal --display-name "Warm Minimal" --keywords "warm,balanced" --scene-tags "warm,clean"

scripts/build_images_only_pptx.py

scripts/build_images_only_pptx.py

Assemble an images-only PPTX by concatenating
```
slide*.png
```
full-screen in filename order.
Use when the user wants “PPT = concatenated images”, with no PPT text boxes.

Dependencies (install if missing):

bash

python3 -m pip install python-pptx

Run:

bash

python3 scripts/build_images_only_pptx.py --images-dir /path/to/slides --out /path/to/deck.pptx

按文件名顺序将
```
slide*.png
```
全屏拼接，生成纯图片PPTX。
用户需要“PPT=拼接图片”、无PPT文本框的场景下使用。

依赖（缺失时安装）：

bash

python3 -m pip install python-pptx

运行：

bash

python3 scripts/build_images_only_pptx.py --images-dir /path/to/slides --out /path/to/deck.pptx

References

参考资料

Read
```
references/prompt_template.md
```
when you need a high-detail prompt skeleton that produces the “rich background infographic” style reliably.
Read
```
references/article_intake_workflow.md
```
when the user gives full article content and expects the skill to infer the deck strategy.
Read
```
references/deck_archetype_routing.md
```
when deciding what overall presentation form best fits the article.
Read
```
references/style-pack-system.md
```
for composition rules and how to add new style packs.
Read
```
references/style-pack-catalog.md
```
for intent-to-style routing and pack selection.
Read
```
references/style-packs/
```
for self-contained style definitions.
Read
```
references/slide_table_normalization.md
```
to normalize underspecified upstream content slide tables into
```
slides_message_plan.md
```
before display and visual planning.
Read
```
references/scene-catalog.md
```
for reusable scene IDs and tags.
Read
```
references/scene-preset-library.md
```
for ready-to-paste scene visual blocks.
Read
```
references/scene-entry-template.md
```
when adding new scene IDs to the catalog.
Read
```
references/scene_library.md
```
to quickly add scene-based / cinematic elements (locations + props) without making the slide messy.
Read
```
references/layout_library.md
```
to choose a layout that matches your text density (so slides stay readable).
Read
```
references/layout_routing_policy.md
```
to auto-route layouts when the user does not explicitly request them.
Read
```
references/shot_mood_library.md
```
for cinematic shot + lighting + time-of-day presets.
Read
```
references/motif_pack.md
```
when you want optional recurring motif guidance for stronger deck cohesion.
Read
```
references/deck_consistency_block.md
```
when you want optional cross-slide consistency locks for light direction, rhythm, and texture behavior.
Read
```
references/typography_spacing_lock.md
```
to prevent tiny text and keep spacing consistent.
Read
```
references/text_fidelity_block.md
```
to reduce text errors and forbid any extra words in any language.
Read
```
references/negative_prompt_block.md
```
to avoid common generator artifacts (watermarks/unspecified text/UI).
Read
```
references/storyboard_library.md
```
to give the whole deck a narrative arc.
Read
```
references/chinese_quote_compression.md
```
only when the required on-slide text is Chinese and long enough to need splitting without paraphrasing.

需要可以稳定生成“丰富背景信息图”风格的高细节提示词框架时，阅读
```
references/prompt_template.md
```
。
用户提供完整文章内容，期望skill推断演示文稿策略时，阅读
```
references/article_intake_workflow.md
```
。
决定哪种整体呈现形式最适配文章时，阅读
```
references/deck_archetype_routing.md
```
。
了解组合规则和如何添加新风格包时，阅读
```
references/style-pack-system.md
```
。
了解意图到风格的路由和包选择时，阅读
```
references/style-pack-catalog.md
```
。
查看自包含的风格定义时，阅读
```
references/style-packs/
```
。
在做展示和视觉规划前，将不明确的上游内容幻灯片表格标准化为
```
slides_message_plan.md
```
时，阅读
```
references/slide_table_normalization.md
```
。
查看可复用的场景ID和标签时，阅读
```
references/scene-catalog.md
```
。
获取可直接粘贴的场景视觉模块时，阅读
```
references/scene-preset-library.md
```
。
向目录中添加新场景ID时，阅读
```
references/scene-entry-template.md
```
。
快速添加基于场景/电影感的元素（位置+道具）且不使幻灯片杂乱时，阅读
```
references/scene_library.md
```
。
选择匹配文本密度的排版（保证幻灯片可读性）时，阅读
```
references/layout_library.md
```
。
用户没有明确要求排版时自动路由排版时，阅读
```
references/layout_routing_policy.md
```
。
获取电影感镜头+光影+时间预设时，阅读
```
references/shot_mood_library.md
```
。
需要可选的重复主题指引增强演示文稿凝聚力时，阅读
```
references/motif_pack.md
```
。
需要可选的跨幻灯片一致性锁定，统一光线方向、节奏和纹理效果时，阅读
```
references/deck_consistency_block.md
```
。
避免文字过小，保持间距一致时，阅读
```
references/typography_spacing_lock.md
```
。
减少文字错误，禁止任何语言的额外文字时，阅读
```
references/text_fidelity_block.md
```
。
避免常见的生成器 artifacts（水印/未指定文字/UI）时，阅读
```
references/negative_prompt_block.md
```
。
为整个演示文稿添加叙事弧光时，阅读
```
references/storyboard_library.md
```
。
仅当要求的幻灯片文字是中文，且长度足够需要拆分而不改写时，阅读
```
references/chinese_quote_compression.md
```
。

infographic-powerpoint-deck

Original

Translation

Infographic PowerPoint Deck

信息图风格PowerPoint演示文稿

Default input model (raw article first)

默认输入模式（优先原始文章）

Artifact model

工件模型

Quick start (default: images-only deck, structured planning)

快速开始（默认：纯图片演示文稿、结构化规划）

Optional structured slide table input

可选结构化幻灯片表格输入

Prompt stack (minimal vs full)

提示词栈（极简版 vs 完整版）

Style-pack system (modular and scalable)

风格包系统（模块化可扩展）

Decision model (keep style, layout, and scene separate)

决策模型（风格、排版、场景分离）

Slide prompt recipe (copy/paste template)

幻灯片提示词模板（可复制粘贴）

Quality bar (and how to iterate)

质量标准（以及迭代方法）

Operating mode (be patient; avoid “probe images”)

运行模式（耐心操作，避免“探测图片”）

Tooling constraints (important)

工具约束（重要）

Bundled scripts

捆绑脚本

`scripts/compose_style_pack_blocks.py`

`scripts/compose_style_pack_blocks.py`

`scripts/create_style_pack.py`

`scripts/create_style_pack.py`

`scripts/build_images_only_pptx.py`

`scripts/build_images_only_pptx.py`

References

参考资料