Communitygithub.com

AiCurv/ADBCommander

Generate a single-file Chinese atelier-style portfolio website (horizontal-row list + hover-reveal thumbnail interaction, artistic Lishu typography via CDN, light/dark theme, hash-routed detail pages). Takes a natural-language description of the studio (name, director, discipline, works) and outputs ONE working .html file. Runs in BOTH Claude Code (writes to ~/Desktop) and GLM / online chat (outputs HTML inline). Trigger phrases include "create a Chinese studio portfolio", "create a Lishu calligraphy portfolio", "lishu-portfolio-gen ...", or when the user drags this skill folder into the chat and asks for an atelier-style site.

What is ADBCommander?

This AI agent skill, portfolio-detail (also known as lishu-portfolio-gen), is designed for developers and designers who want to quickly generate a static, single-file HTML portfolio website with a refined, Chinese atelier aesthetic. It specializes in horizontal text rows with hover-reveal thumbnails, large artistic Lishu calligraphy (Alimama DaoLiTi font via CDN), a light/dark theme, and hash-routed detail pages. The skill can be triggered by natural language descriptions of a studio (name, director, discipline, works) or by specific phrases like "create a Lishu portfolio" or "lishu-portfolio-gen ...". It operates in two modes: Mode A for local environments like Claude Code with file system access (saving to ~/Desktop), and Mode B for online chats like GLM without file system access (outputting the HTML inline). The tool reads a template file, user-provided or AI-generated content, and optionally images, then swaps in data for up to 8 works (with JSON data, prose, specs, gallery, and credits). It also includes crucial rules for JSON formatting (full-width quotes), image sources (picsum IDs), slug consistency, and preserving the existing CSS/JS system.

Works with✓Claude Code~Codex CLI~Cursor
npx skills add https://github.com/AiCurv/ADBCommander/tree/HEAD/skills/design/design-templates/portfolio-detail

Ask in your favorite AI

Open a new chat with this agent skill pre-loaded.

Documentation

portfolio-detail — Chinese Atelier / Lishu Calligraphy Portfolio (with Secondary Detail Pages)

One-Line Characteristic

Horizontal text rows + hover to reveal large image on the right + Alimama DaoLiTi artistic large characters + black-and-white pure colors + light/dark theme + hash-routed detail pages + list/gallery dual views.

Differences from Other Portfolio Skills

skillcore modebest for
portfolio-geninfinite drag canvas + tile layoutPhotographers / graphic designers showcasing samples
portfolio-museum-genhero carousel + large cards + waterfall + detail pages (film curation feel)Video/film/installation artists
industrial-archive-genblack-framed flip cover + sidebar archive + Nordic industrialPhysical products / objects / hardware
lishu-portfolio-genHorizontal text rows + hover reveals right thumbnail + Lishu large characters + restrained Chinese aestheticsBookbinding / typography / craft / Chinese ateliers

If the user's project is about "books / typography / exhibitions / still life / Chinese herbs / tea / pottery / wood / paper / incense" and requires a restrained, elegant style — use this one.

When to Trigger

The user says:

  • "Create a Chinese / Eastern / classical style portfolio"
  • "Create a Lishu-style website / studio homepage"
  • "Create a white-atelier style portfolio" / "Create a horizontal list website like Gurgel D'Alfonso"
  • "lishu-portfolio-gen ..."

And provides:

  • Studio description (Required): name, director, city, discipline, titles and stories for several works
  • Work images (Optional): URL / uploaded file / path. It works without them (uses picsum placeholders)

If the user does not provide work titles at all → ask at least which works need to be displayed. If the user only provides titles but no details → write them yourself: generate 1 lead sentence + 3 prose paragraphs + 6-8 spec lines + 3-6 picsum images + several credits for each work. The tone should match the work type (book → bookbinding / typography → font / object → still life / exhibition → exhibition visuals).

Two Operation Modes

Mode A · Claude Code / Local (with file system)

  • Use Bash to list folders, sips to probe image dimensions
  • Output location: ~/Desktop/{slug}-portfolio.html (single file, all images via URL)
  • If user provides local images: copy them to ~/Desktop/{slug}-portfolio-images/ and update src to relative paths

Mode B · GLM / Online Chat (no file system)

  • User drags in a zip archive, after extraction they see 4 files
  • Only output one complete HTML string, wrapped in a ```html code block
  • If user uploads images but does not provide URLs → use https://picsum.photos/id/{id}/{w}/{h} placeholders (use known existing IDs)
  • Do not attempt to access local files or suggest command-line operations

Decision: If you cannot Read to see local images → default to Mode B.

File Division

FileRoleHow to use
reference.htmlComplete finished reference — BAIWU atelier with 6 worksLook at: wording of list items, prose tone, spec vocabulary, JSON field density
template.htmlEmpty skeleton — same CSS/JS/routing system, 2 examples + 20 SWAP markersModify based on it, output final HTML
README.mdUser-facing instructionsNot needed by the model
SKILL.mdWorkflow instructions for the model (this file)Must read

Core rule: All modifications based on template.html; learn how to write (tone, word count, tag wording) from reference.html, but do not copy its specific content.

Work Steps

1. Read the Two HTML Files

  • Read reference.html first to understand "what the complete version looks like" (especially the field density of the works-data JSON and prose tone)
  • Then read template.html to find the 20 SWAP locations

Do not modify JS / CSS / detail page routing system / font imports. SWAPs are only in HTML text and JSON data.

2. Parse User Input and Extract Information

Extract:

  • ATELIER_NAME_ZH — Chinese studio name (2-4 characters best, e.g. "白屋" "素园" "墨记" "青庐" "拙轩")
  • PINYIN — Pinyin (5-6 English letters, all uppercase: BAIWU / SUYUAN / MOJI)
  • CHAR — Single character for the logo (usually the first character of the studio name: 白 / 素 / 墨 / 青 / 拙)
  • CITY — City
  • DIRECTOR_ZH / DIRECTOR_EN — Director's name + pinyin
  • DISCIPLINE — 3 discipline words ("装帧 · 字体 · 静物")
  • TAGLINE — One-line positioning (20-40 characters, kai-style temperament)
  • YEAR_FOUNDED — Founding year (can use Gan-Zhi: 乙未冬 / 甲辰春)

For each work extract:

  • slug — URL ID, lowercase English with hyphens (kong-shan-ji / wu-ming-shu)
  • title_zh — Chinese title (2-5 characters best, more concise = more character)
  • title_en — English subtitle (book-title style: An Anthology of Empty Mountains)
  • year — Year or Gan-Zhi year (甲辰 / 2024 / 2025 / 乙巳 / 2025)
  • tag — Top category, wrapped in square brackets, e.g. [ 诗集 · 装帧设计 ] / [ 字体设计 · 中草药图谱 ]
  • cover — Detail page large image URL
  • list row location — City, Year · Craft/Spec (appears on hover)
  • meta — 4 key-value pairs (4-column summary on the detail page)
  • lead — 30-70 character opening large text, can use <em>keyword</em> for emphasis
  • prose — 3-4 paragraphs, each 80-150 characters
  • specs — 6-10 row spec table (format/paper/binding/typeface/print run/printing/craft/price etc.)
  • gallery — 3-6 images, layout one of three: g-wide / g-half / g-third
  • credits — 4-7 pairs of role/name

If user only provides one sentence and no story → create it yourself: write lead and prose that match the work's title and type. Reference the tone of reference.html — no marketing language, have a "museum wall-text + studio journal" feel (restrained, specific, slow, with traces of handcraft).

3. List Works + Check Dimensions (Mode A only)

ls -la "{works_path}"
sips -g pixelWidth -g pixelHeight "{works_path}/{file}" 2>/dev/null

Obtain (filename, w, h, aspect) for each image. Prefer vertical images for list thumbnails (4:3 container, vertical images look better).

4. Replace 20 SWAPs

SWAPLocationContent
1<title>{ATELIER_NAME} {PINYIN} — {DIRECTOR} Portfolio / Studio
2meta descriptionOne-line site description
3curtain single characterFirst character of studio name (e.g. "白")
4logo SVG inner single characterSame as SWAP-3
5logo SVG inner pinyin{PINYIN} uppercase
6main navigation liUsually 3 items: Works / About / Contact
7brand-block Chinese name{ATELIER_NAME} with space between characters (e.g. 白&nbsp;&nbsp;屋)
8brand-block English{PINYIN} Studio · {CITY}
9taglineOne-line positioning in kai style, 20-40 characters, can use <br> for line break
10list items N piecesEach <a class="item" href="#work/{slug}"> contains num/project-name/en-name/location/photo-thumb
11gallery items N piecesSame N works, more compact grid — slug must match list
12About title关&nbsp;于&nbsp;{name}<span class="en">About</span>
13About paragraphs1-3 paragraphs in kai style (background / director / design philosophy), each 60-120 characters
14About facts table4-6 items li (Founded/Location/Discipline/Director/Hours)
15Contact title来&nbsp;信<span class="en">Get in Touch</span>
16Contact paragraphs1-2 paragraphs inviting letters + reply cycle commitment
17Contact facts table4-6 items (E-mail/Phone/Address/Instagram/WeChat Official Account)
18rotating creditSite by {WHO} (usually the studio name)
19footer three blocks© / ICP / social links
20works-data JSONComplete detail data for N works (critical!)

5. Write works-data JSON (Most Critical Step)

Each work must have one entry in the JSON array. The slug must exactly match the #work/SLUG in the list/gallery card href.

JSON fields:

{
  "slug": "kong-shan-ji",
  "title_zh": "空山集",
  "title_en": "An Anthology of Empty Mountains",
  "year": "甲辰 / 2024",
  "tag": "[ 诗集 · 装帧设计 ]",
  "cover": "https://picsum.photos/id/143/1800/1100",
  "meta": [
    {"label": "Type / 类型", "value": "诗集装帧"},
    {"label": "Year / 年份", "value": "2024"},
    {"label": "Edition / 印量", "value": "首印 320 本"},
    {"label": "Format / 开本", "value": "130 × 195 毫米"}
  ],
  "lead": "三十六首关于山的诗。我们想做一本——读者翻开后,先听见<em>风声</em>,再读见字的书。",
  "prose": [
    "Paragraph 1: Source of poems / design context. 80-150 characters.",
    "Paragraph 2: Key decisions during production — typeface, paper, binding choices.",
    "Paragraph 3: Final presentation / feeling left for the reader."
  ],
  "specs": [
    ["Format / 开本", "130 × 195 毫米"],
    ["Pages / 页数", "144 页"],
    ["Paper / 纸张", "本色棉纸 90gsm"],
    ["Binding / 装订", "线装四目"],
    ["Typeface / 字体", "老宋(标题)· 楷体(正文)"],
    ["Printing / 印刷", "单色凸印"],
    ["Edition / 印量", "首印 320 本"],
    ["Price / 定价", "¥ 388"]
  ],
  "gallery": [
    {"src": "...", "layout": "g-wide", "caption": "封面"},
    {"src": "...", "layout": "g-half"},
    {"src": "...", "layout": "g-half"},
    {"src": "...", "layout": "g-third"},
    {"src": "...", "layout": "g-third"},
    {"src": "...", "layout": "g-third"}
  ],
  "credits": [
    ["Design / 设计", "周墨白"],
    ["Hand Binding / 手工装订", "陈漱石(杭州·订书坊)"]
  ]
}

JSON Gotchas ⚠️:

  • Quotes in Chinese prose must use full-width 「...」 or "..." (U+201C / U+201D), not ASCII " (will break JSON)
  • Do not use real newlines inside JSON strings; to indicate paragraph breaks, add another item in the prose array
  • <em>word</em> in lead is written directly as HTML tags — rendered as innerHTML
  • Gallery layout must be one of three: g-wide (full width) / g-half (half width) / g-third (one third)

6. Image Source Selection

  • picsum placeholder: Use id/X format instead of seed/X (more stable). Known good IDs: 0, 20, 40, 51, 64, 76, 88, 91, 96, 119, 121, 130, 143, 145, 158, 164, 165, 175, 177, 180, 187, 195, 200
  • Use adjacent IDs for cover/gallery of the same work to create a consistent color tone
  • Do not use ?grayscale parameter (users expect images to be in color)

7. Chinese Studio Name Inspiration (if user doesn't provide one)

If the user asks you to "create a name yourself" → choose 2 characters + ancient style:

  • Object category: 白屋 / 素园 / 墨记 / 青庐 / 拙轩 / 听雨 / 一勺 / 半亩
  • Character/calligraphy category: 素白 / 子衿 / 砚池 / 半山 / 寻常 / 浮云
  • Avoid functional words like "工作室 / 设计 / Studio", let the name itself be a poem

8. Video Support

If a work's cover is .mp4 / .webm / .mov, the detail page will automatically use <video autoplay muted loop> instead of <img>. Just provide the video path in the cover URL.

9. Reporting

Mode A: One-sentence report.

Generated ~/Desktop/baiwu-portfolio.html — 6 works, hover over rows to view thumbnails, click to enter detail pages. Light/Dark toggle at bottom left, List/Gallery toggle at top center.

Mode B: Provide the complete HTML code block + one line of explanation.

Save the above snippet as index.html and double-click to open it. Hover your mouse over any work name to see the thumbnail appear on the right, click to jump to the detail page. Replace picsum URLs with your own image addresses.

Mandatory Items (Cannot Change)

  • Font import: @font-face loading Alimama DaoLiTi woff2 URL (do not change the font source)
  • CSS variable system: light/dark dual themes
  • All JS: theme toggle, view mode toggle, hover-row interaction, rotating credit, detail page hash routing applyRoute(), renderDetail()
  • <script id="works-data"> node must be preserved
  • <main id="detail-root"> must be preserved (detail page mount point)
  • body[data-view="home/detail"] and body[data-mode="list/gallery"] state system
  • #logo / #main-menu / .theme-toggle / #project-view / #svp-credit five fixed-position elements

Quick Error Checklist

  • Slug mismatch: list item #work/kong-shan-ji ↔ gallery item #work/kong-shan-ji ↔ JSON "slug": "kong-shan-ji" — all three must be identical
  • JSON quotes: Use full-width 「」 or "..." for Chinese content quotes, not ASCII "
  • Too many gallery items: Maximum 6 images per work, exceeding this breaks the layout
  • Prose too long: Each paragraph 80-150 characters, readability drops if over 200 characters
  • List / Gallery out of sync: Must contain the same set of works (one-to-one correspondence)
  • Picsum seed: Use id/X instead of seed/X to avoid random IDs hitting deleted images
  • Works < 3: It works, minimum 2 works
  • Works > 8: Suggest splitting into two batches, too many list rows make the page long
  • First character drop cap: .detail__prose p:first-of-type::first-letter uses DaoLiTi font — the first paragraph must start with Chinese characters (not with <em> or English)
  • Removed picsum color: Images should remain in color, do not add filter: grayscale() in CSS

Individual skills in this repo

This repo contains 7 individual skills — each has its own dedicated page.

AiCurv/ADBCommander

Premium calm-tech SaaS landing template — bilingual (CN/EN) literary voice over a dark forest-night palette with warm gold accent, Cormorant + Noto Serif SC, Three.js procedural forest hero, station-based scrolling narrative, custom cursor. Built for ambient-AI / creative-tool / philosophical SaaS, not loud commercial launch.

AiCurv/ADBCommander

Pro-camera HUD mockup template — full-viewport black canvas wrapping a real video with monospace overlays (corner brackets, top status bar, side AF/exposure columns, bottom timecode + aperture/shutter/ISO row, vertical 8-filter wheel, animated reticle). For cinematographer portfolios, camera-app demos, LUT pack previews, filter-look auditions. NOT a TikTok/Reels mockup, NOT a storyboard, NOT a shot list, NOT a script document. The chrome wraps real footage — never a div placeholder.

AiCurv/ADBCommander

Premium pre-launch waitlist page template — dark charcoal canvas with subtle vignette, drifting portrait video-thumbnail cards, fade-up cascades, glass nav, live-updating join counter, single email form with success/error states. Built for creative-tool / AI / niche-SaaS invite-only launches with hushed exclusive tone, not loud growth-marketing.

AiCurv/ADBCommander

Route design-related HTML artifact tasks to the right artifact skill, reference, design system generation, or export skill.

AiCurv/ADBCommander

Implement text-to-speech (TTS) capabilities using the z-ai-web-dev-sdk. Use this skill when the user needs to convert text into natural-sounding speech, create audio content, build voice-enabled applications, or generate spoken audio files. Supports multiple voices, adjustable speed, and various audio formats.

AiCurv/ADBCommander

Implement AI-powered video generation capabilities using the z-ai-web-dev-sdk. Use this skill when the user needs to generate videos from text prompts or images, create video content programmatically, or build applications that produce video outputs. Supports asynchronous task management with status polling and result retrieval.

AiCurv/ADBCommander

Implement specialized video understanding capabilities using the z-ai-web-dev-sdk. Use this skill when the user needs to analyze video content, understand motion and temporal sequences, extract information from video frames, describe video scenes, or perform video-based AI analysis. Optimized for MP4, AVI, MOV, and other common video formats.

Related Skills