Home / All Skills / Design & Motion / shan-ze-school
English (EN)简体中文 (ZH)日本語 (JA)
Design & Motion · vibeshotclub/vsc-skills

shan-ze-school

Generate or refine prompts for 杉泽流派 / Shan Ze school-inspired new-oriental mythic ink-and-color fantasy images, especially when the user asks for 山海经、神怪、异兽、妖怪、国风神话、工笔水墨奇幻、东方幻想生物插画, or wants an existing prompt rewritten into this visual language. Use for original image-generation prompts, prompt rewrites, style transfer phrasing, and concise style analysis that translates any living-artist reference into broader art-historical visual grammar.

L1 Basic · Source registration onlyV1 Commit-PinnedDual-pass Audit
Author vibeshotclubGitHub vibeshotclub/vsc-skills ↗Stars 832License 仓库未提供 LICENSE 文件（find 全仓无 LICENSE*；GitHub API license 字段为 null）Commit fcee81beb6
Control levels describe skill capabilities and suggested control measures; they do not represent quality or safety certifications, nor do they dictate usability. Usage decisions rest entirely with users and administrators. Different agent hosts apply varying constraints; do not assume hosts provide automatic protection. Third-party skill copyrights belong to their original authors. Use at your own risk.

1How It Works & Mechanism

The core mechanism is a translation rule mapping "living artist name → art-historical visual grammar": living artist names must never appear in the final prompt.

shan-ze-school/SKILL.md
Translate requests for "Shan Ze / 杉泽 style" into an original art-historical visual grammar. Do not put a living artist's name in the final image prompt.
A sequence of replacement descriptors is immediately provided: new-oriental mythic ink illustration, gongbi linework, xieyi ink atmosphere, shan-hai-jing bestiary, mineral pigments, silk-paper texture, smoke-cloud negative space.

Prompts are built on a four-tier structure (Structure/Color/Atmosphere/Finish), with each tier populated by concrete terminology rather than stacked adjectives.

shan-ze-school/SKILL.md
- **Structure**: gongbi baimiao contour, iron-wire line, hairline tendrils, anatomical hybridization.
The other three tiers are Color (ink gray, cinnabar, rouge red, malachite green, azurite blue, ochre, pale shell white), Atmosphere (ink-wash smoke, cloud-reserve negative space…), and Finish (silk or xuan-paper grain…).

Drift prevention relies on a fixed set of negative constraints that explicitly exclude alien styles such as Western oil painting, Japanese cel-shading, 3D rendering, and sci-fi mecha.

shan-ze-school/SKILL.md
4. Add negative constraints to prevent drift: no Western oil painting, no anime cel shading, no 3D render, no hard sci-fi armor, no glossy digital airbrush, no neon cyberpunk.
The end of the Prompt Template includes a separate Negative prompt line enforcing the same exclusions.

In-depth vocabulary is externalized into an art-historical grammar reference, loaded on demand rather than read entirely by default.

shan-ze-school/SKILL.md
For deeper vocabulary, read [references/style-grammar.md](references/style-grammar.md). Load it when the user asks for a very detailed prompt, variations, art direction, or style analysis.
This reference contains eight sections: Art-Historical Base / Line And Brushwork / Color System / Light And Space / Creature Design / Composition Patterns / Prompt Add-Ons / Negative Constraints.

Four output modes are available, and the presence of image generation tools determines whether an image is actually generated; without tools, it falls back to a ready-to-use prompt.

shan-ze-school/SKILL.md
If the user asks to generate an image, generate the image when an image tool is available. If direct image generation is not available, provide a ready-to-use prompt.
The Output Modes section lists four variants: Direct image generation / Ready-to-use prompt / Prompt rewrite / Mini style note.

When rewriting existing prompts, only modify the style tier while preserving the user's core subject, preventing the subject from being overridden by style grammar.

shan-ze-school/SKILL.md
5. If the user supplied a draft prompt, preserve the core subject and only rewrite the style layer.
The Prompt rewrite mode is described as: "keep the user's subject, simplify weak adjectives, and replace artist-name shorthand with material, brushwork, composition, and negative constraints".

Compositions are enumerated as selectable options rather than free-form descriptions to ensure consistent output: floating creature, handscroll chase, screen painting under flora, and bestiary plate.

shan-ze-school/SKILL.md
- bestiary plate with seal and minimal inscription
The same section includes floating creature in blank paper space, horizontal handscroll pursuit scene, and screen-painting tableau under flowering canopy.

Tool boundaries are explicitly defined in SKILL.md: out-of-scope directions are deferred to other skills rather than handled forcefully.

shan-ze-school/SKILL.md
- The user needs a broad Chinese illustration taxonomy rather than a narrow prompt language. In those cases, prefer `$中国插画`.
The Do Not Use section also excludes art criticism, art-historical attribution, academic classification, as well as contemporary graphic design, anime, realistic concept art, and Western fantasy.

2Core Capabilities

01Dual-purpose for generation/rewriting: directly generates images when image tools are available, or provides ready-to-use prompts when tools are absent
02Full copy-paste template: subject placeholder + seven categories of gongbi/ink terminology + four tiers of structure, color, atmosphere, and finish + Negative prompt
03Four default composition types (Blank-field floating / Handscroll chase / Screen under flora / Bestiary plate)
04Art-historical vocabulary base (baimiao / iron-wire line / gossamer line / nail-head mouse-tail line, Dunhuang mineral pigments, negative space and ink-cloud atmosphere)
05Compositional motif library (Blank-field apparition / Handscroll chase / Flower canopy elegy / Bestiary plate / Ritual encounter)
06Mini style note mode: when asked "why does this resemble the Shan Ze school", only explain visual characteristics that translate directly into prompts
07Disciplined variant generation: modify only one axis at a time (composition / pigment emphasis / atmosphere / creature anatomy)

4Recommended Governance & Evidence

Suggested Controls for Safe Operation
For Agent Runtime
Can be auto-loaded
For Administrators
Register source and version
Evidence & Triggers (File Paths & Excerpts)
  • SKILL.md: Pure prompt instructions and rule text, without script execution or external impacts
Audit Notes: No scripts, network access, credentials, or disk writes; repository contains only SKILL.md, README.md, references/style-grammar.md, and agents/openai.yaml. The design proactively mitigates risks: living artist names are replaced with art-historical visual grammar, and a fixed Negative prompt excludes foreign styles. Key considerations are that actual image generation depends on host tools and rules rely on model execution.

5Second-Pass Verification

  • [ok] 外部资源/工具条目（本 skill 无） — 仓库级 https?:// 扫描中，本目录零命中；SKILL.md 未要求任何 CLI/API/包。『实际出图』依赖宿主自带图像工具，属宿主能力而非 skill 依赖。
  • [ok] 是否漏掉隐藏脚本或模板 — 文件清单仅 4 个明文文件；无 scripts/、无 assets/、无 base64/二进制块。
  • [ok] art-historical 转译规则是否真在指令里（而非仅 README 宣传） — SKILL.md Core Rule 段与 Prompt Template 均落地该规则，且 style-grammar.md 的 Negative Constraints 首条即『no direct living-artist name』。
  • [ok] 风格语法库与 SKILL.md 是否重复/脱节 — SKILL.md 四层结构（Structure/Color/Atmosphere/Finish）与 style-grammar.md 的 Line And Brushwork / Color System / Light And Space 三节术语一致，参考文档是扩写而非另立体系。
  • [ok] 路由边界声明是否与 vsc 目录登记一致 — SKILL.md vsc-distinction 指向「广泛探索多种风格优先风格探索」即 rare-style-explorer，与 rare-style-explorer 的 vsc-distinction 反向指向一致。
  • [ok] 安全结论反例搜索 — 逐 token 扫描无网络/环境变量/写盘/子进程；无混淆内容；SKILL.md 未授权任何对外发布或安装动作。

6Verdict & Scenarios

  • Safely translates "living artist style" requests into art-historical visual grammar while preserving the recognizable gongbi aesthetic of oriental mythic beasts.
  • Ready-to-use prompt template: four-tier structure + fixed single-line Negative prompt for easy copying and high output stability.
  • Highly professional terminology base: terms like baimiao, iron-wire line, gossamer line, nail-head mouse-tail, negative space, and mist can be copied directly into prompts.
  • Clear boundaries: avoids art criticism, does not overlap with broad Chinese illustration demands, and stays within its routing scope.
    • Best for: Suitable for creating and rewriting image generation prompts for Shan Hai Jing, mythical spirits, fantastical beasts, and traditional Chinese mythic themes; ideal for compliance-driven rewriting that converts "living artist style" requests into unnamed visual grammar; also fits scenarios needing a concise style breakdown to integrate into other prompt workflows.
    Not for: Not suitable for art criticism, style attribution, or academic classification (explicitly excluded in SKILL.md); not suitable for contemporary graphic design, anime, realistic concept art, or Western fantasy; use `$中国插画` when a broad-spectrum Chinese illustration taxonomy is required; use rare-style-explorer for broad multi-style exploration.
    Install Agent-Ready
    ① Verified Mirror 2026-09-15
    Option A · Direct DownloadDownload shan-ze-school.tar.gz
    sha256: 265ceda4f4bb7e83…
    Option B · Copy Guide for Agent
    install.json
    Agents automatically download and verify mirrors from this JSON guide.
    ② Upstream GitHub
    GitHub Repo ↗
    Pinned to commit fcee81beb6.