Magazine Web Ppt

Magazine Web Ppt

What this skill does

Generates a single-file HTML horizontal-swipe deck with a visual tone of:

• A electronic magazine + electronic ink hybrid style
• A WebGL fluid / contour / dispersion background (visible on hero pages)
• Serif headlines (Noto Serif SC + Playfair Display) + sans-serif body (Noto Sans SC + Inter) + monospace metadata (IBM Plex Mono)
• Lucide line icons (no emoji)
• Horizontal left/right paging (keyboard ← →, scroll wheel, touch swipe, bottom dots, ESC index)
• Smooth theme interpolation: when you flip to a hero page, the colors and shader transition softly

This skill's aesthetic is not "corporate PPT," nor "consumer-internet UI": it looks like Monocle magazine after code was stitched onto it.

When to use

Suitable scenarios:

• Offline talks / internal industry talks / private salons
• AI new-product launches / demo day
• Talks with a strong personal style
• A web version of slides that you "make once, with no paging tool needed"

Unsuitable scenarios:

• Long stretches of tabular data, stacked charts (use a regular PPT)
• Training courseware (information density too low)
• Multi-person collaborative editing (this is static HTML)

Workflow

Step 0 · Infer a direction; ask only when comparison is requested

Infer one of the 5 magazine directions from the brief, audience, content, and known context. Each direction packages its theme color, recommended layouts, chrome style, and slide count. Use the closest fit and continue without asking.

Only when the user explicitly asks to see or compare direction options, open references/styles.md and present the 1-line summaries in one direction-cards question form:

1. Monocle Editorial · International magazine style ✦ default
2. WIRED Tech · Data + engineering
3. Kinfolk Slow · Slow living / humanist
4. Domus Architectural · Architecture / sense of space
5. Lab / Reference · Academic + craft manual

If the user says "I don't know, you recommend" or provides no direction cue: default to Monocle Editorial without asking, because it has the lowest failure probability. If the user mentions "AI / benchmark / technical launch": choose WIRED; "reading / private / social circle": choose Kinfolk; "design / architecture / portfolio": choose Domus; "research / academic / methodology": choose Lab.

After picking a direction, create or update 项目记录.md (Project Record) in the project folder, with the first line clearly stating direction + theme color + audience + duration (see the template at the end of styles.md). Do not change direction at any point: switching midway = everything before is wasted.

Step 1 · Resolve intent

If the user has already given a complete outline + images, you can skip straight to Step 2.

Infer the checklist below from the brief, project metadata, supplied content, and conversation. If a missing answer would materially change the deck and no safe default exists, ask only those unresolved fields in one consolidated question form. Do not ask the checklist one by one.

6-question clarifying checklist

Direction and theme color are inferred in Step 0 unless the user explicitly asks to compare options.

Outline assistance (if the user has no outline)

Use the "narrative arc" template to build the skeleton, then fill in content:

Hook       → 1 page   : throw out a contrast / question / hard number that makes people stop
Context    → 1-2 pages : explain the background / who you are / why this topic
Core       → 3-5 pages : the core content, interleaving Layouts 4/5/6/9/10
Shift      → 1 page    : break the expectation / propose a new view
Takeaway   → 1-2 pages : a punchline / a suspenseful question / a call to action

Narrative arc + page-count plan + theme rhythm table (see layouts.md); align all three tables before moving to Step 2.

It's best to save the outline as 项目记录.md (Project Record) or 大纲-v1.md (Outline v1) for later iteration.

Image convention (tell the user)

Make these clear to the user before starting:

• Folder location: under 项目/XXX/ppt/images/ (sibling to index.html)
• Naming convention: {page number}-{semantic}.{ext}, e.g. 01-cover.jpg / 03-figma.jpg / 05-dashboard.png
  • Pad the page number with zeros for easy sorting
  • Use English for the semantic part, short, specific, matching the content
• Spec recommendations:
  • Each image ≥ 1600px wide (avoid blur on large screens)
  • JPG for photos/screenshots, PNG for transparent UI/charts
  • Keep total size under 10MB (affects paging smoothness)
• How to replace: same-name overwrite is the most stable (no need to change the path in the HTML); if the filename changes, remember to globally search images/oldname and change it to the new name
• What if there are no images: align with the user; you can generate the structure with placeholder color blocks first and fill in the images later, but tell them that image-text mix pages like layouts 4/5/10 can't have their visuals verified without images

Step 2 · Copy the template

Copy assets/template.html to the target location (usually 项目/XXX/ppt/index.html), and create an images/ folder as a sibling, ready to receive images.

mkdir -p "项目/XXX/ppt/images"
cp "<SKILL_ROOT>/assets/template.html" "项目/XXX/ppt/index.html"

template.html is a complete, runnable file: CSS, WebGL shader, paging JS, font/icon CDNs are all preset; only inside <main id="deck"> are there 3 example slides (cover, act divider, blank filler page).

2.1 · Placeholders you must change (easy to miss)

Right after copying, change the placeholders below, or the browser tab will show awkward text like "[必填] replace with the PPT title":

The first thing to do every time you finish copying template.html: grep for "[必填]" to confirm everything is replaced.

2.2 · Choose the theme color (5 presets · no customization allowed)

This skill only allows picking one of 5 carefully tuned presets; it does not accept user-defined hex values: a wrong color pairing makes the visuals ugly in an instant, and protecting the aesthetic matters more than offering freedom.

How to do it:

1. Recommend one based on the content theme, or just ask the user which one to pick
2. Open references/themes.md and find the :root block for the corresponding theme
3. Replace as a whole the lines marked with the "theme color" comment in the opening :root{ block of assets/template.html (the copied version): --ink / --ink-rgb / --paper / --paper-rgb / --paper-tint / --ink-tint
4. All other CSS flows through var(--...), no other changes needed

Hard rules:

• A deck uses only one theme, don't change colors midway
• Don't accept an arbitrary hex value from the user: gently decline and show the 5 presets to pick from
• Don't mix and match (e.g. take ink from Ink Classic and paper from Dune): it clashes completely

Step 3 · Fill in content

3.0 · Pre-check: class names must be defined in template.html (most important)

This is the source of all generation problems. The layouts.md skeletons use many class names (h-hero / h-xl / stat-card / pipeline / grid-2-7-5, etc.); if assets/template.html's <style> has no corresponding definition, the browser falls back to default styles: the big title turns sans-serif, the data cards squeeze into a clump, the pipeline mushes into one line, the images pile up at the bottom of the page.

Before writing any slide code:

1. First Read assets/template.html (at least through the end of the <style> block)
2. Against layouts.md's Pre-flight list, confirm that every class you're going to use exists in the <style>
3. If a class is missing: add it to template.html's <style>, don't rewrite it inline on every slide
4. template.html is the single source of class names: don't invent new class names; if you need a custom one, use style="..." inline

Classes commonly missed (must be confirmed to exist in advance): h-hero / h-xl / h-sub / h-md / lead / kicker / meta-row / stat-card / stat-label / stat-nb / stat-unit / stat-note / pipeline-section / pipeline-label / pipeline / step / step-nb / step-title / step-desc / grid-2-7-5 / grid-2-6-6 / grid-2-8-4 / grid-3-3 / grid-6 / grid-3 / grid-4 / frame / frame-img / img-cap / callout / callout-src / chrome / foot