Kouzui-HUANG

專為 Suno 與 Udio 設計的 AI 音樂提示詞（Prompt/Tags）生成助手。當使用者提出抽象的情緒、畫面或曲風融合需求（如：「古風結合賽博龐克」）時，請觸發此技能。它會將模糊的靈感精準轉譯為專業的音樂流派、樂器、混音與結構標籤，大幅提升 AI 音樂生成的品質與準確度。

中文简介正在准备中。当前显示原文说明。

Hollywood Screenwriter & Character Architect (Oscar-grade, with a sociologist's and psychologist's acuity). Reverse-engineers living, dramatically charged characters from fragmentary clues (a line of dialogue, an accessory, or a single photo's composition and lighting) and generates THREE maximally differentiated, internally self-consistent character profile documents in Traditional Chinese — covering physical/social attributes, family background and trauma, MBTI/zodiac/blood-type coherence, speech habits & micro-expressions, plus a cinematic executive summary. Trigger when the user mentions "角色設定", "人格設定", "人物塑造", "人設", "角色檔案", "建立角色", "character profile", "character design", "persona creation", "character building", or wants to create/develop/expand a fictional character's settings or backstory. Do NOT trigger for image-generation prompt requests (portraits, character sheets, moe characters) — those belong to prompt-master-portrait / prompt-master-character-sheet / prompt-master-otaku-*.

中文简介正在准备中。当前显示原文说明。

Chief Character Costume & Visual Design Director. Generates logically consistent, narratively-driven wardrobe designs for fictional characters based on their personality, background, mood, occasion, and climate. Outputs bilingual (EN/ZH) professional fashion descriptions using precise industry terminology. Use this skill when the user mentions "服裝設計", "穿搭", "角色穿什麼", "outfit", "costume", "wardrobe design", or asks what a character would wear in a given scene or context.

中文简介正在准备中。当前显示原文说明。

Facebook Viral Post Architect (社群爆文架構師) — turns a marketing need into a publish-ready Traditional Chinese Facebook post engineered against the platform's actual ranking signals (share weight, follow-conversion, dwell time, negative-signal avoidance). Delivers ONE main post (mobile-formatted, with its first-comment link) + THREE alternative hooks on different formulas + a signal-design rationale. Picks and declares one strategic axis — 分享型 (reach) / 追蹤型 (follower growth) / 轉單型 (conversion). Asks for raw material only when the request lacks the concrete fuel a viral post needs. Use when the user mentions 「FB貼文」,「FB文案」,「臉書貼文」,「臉書文案」,「粉專貼文」,「社群貼文」,「爆文」,「寫貼文」,「貼文企劃」,「貼文腳本」, "FB post", "Facebook post", "Facebook copy", "social post", "viral post", or wants a marketing message/product/service/personal brand turned into a high-reach Facebook post. It ONLY writes post TEXT — it never generates images or video, never calls any generation API, and never publishes anything. Do NOT trigger for — 短影音腳本 / short-video scripts (use prompt-master-short-video), 歌詞 / lyrics (use prompt-master-lyrics), image prompts of any kind (use the relevant prompt-master-* skill), or social graphic/visual design (use social-media-visual-designer).

中文简介正在准备中。当前显示原文说明。

Executable tool skill that ACTUALLY CALLS the GMI Cloud `gemini-3-pro-image-preview` API to generate or edit real images — it runs a bundled Python script (submit → poll → download), it is NOT a prompt writer. Activation is by MODEL NAME ONLY. Trigger EXCLUSIVELY when the user explicitly names the model: "gemini 3 pro image", "gemini-3-pro-image", "gemini 3 pro image preview", "nanobanana pro", "nano banana pro", "奈米香蕉pro", or "奈米香蕉 pro". CRITICAL — DO NOT trigger on generic image-related wording: "圖片生成", "圖片編輯", "image generation", "image editing", "九宮格 / nine-panel", "動漫化 / anime", "character sheet / 三視圖", "product photo / 產品圖", "makeup / 彩妝", "storyboard / 分鏡", "reverse / 逆向", etc. — those describe IMAGE PROMPT tasks handled by the many other prompt-writing skills in this project and are UNRELATED to this skill. Also do NOT trigger on bare "nano banana" / "奈米香蕉" WITHOUT "pro" (that is a different model). Only the explicit model name above activates this skill; if the user merely describes a picture but never names this model, do NOT use this skill.

中文简介正在准备中。当前显示原文说明。

Automates the process of initializing a local Git repository, handling remote conflicts, and pushing code to GitHub. Trigger when the user mentions "push", "上傳github", "push to github", or asks to upload/push their code to an upstream repository.

中文简介正在准备中。当前显示原文说明。

Generate or edit images with OpenAI's gpt-image-2 via GMI Cloud's async request-queue API (console.gmicloud.ai). One dual-mode script: with NO input image it runs `gpt-image-2-generate` (text-to-image); with one or more input images it runs `gpt-image-2-edit` (one request per image). Submits the job, polls until done, and downloads every result image to the shared project-root `outputs/` folder. Defaults: quality=medium, output_format=png, size=1920x1088, n=2. API key is read from a `.gmi_api_key` file in the skill folder (also accepts `--api-key` or `$GMI_API_KEY`). ACTIVATION IS BY EXPLICIT MODEL NAME ONLY. This skill ACTUALLY GENERATES / EDITS real images and spends API credits, so TRIGGER ONLY when the user explicitly names this model / provider — "gpt-image-2", "gpt image 2", "GMI", "GMI Cloud" — e.g. "用 gpt-image-2 生圖", "用 gpt-image-2 改圖", "GMI 生圖", "GMI Cloud 生圖", "跑 gpt-image-2", "gpt image 2 編輯", "gpt image 2 生成". DO NOT TRIGGER on generic image wording that does NOT name gpt-image-2 / GMI — e.g. "編輯圖片", "圖片編輯", "載入圖片編輯", "圖片編輯技能", "改圖", "修圖", "編輯這張圖", "轉為黑白線稿", "圖片生成", "生圖", "image editing", "edit image", "image generation". Those are PROMPT-WRITING requests, NOT generation: • editing an image / a hypothetical visual change → use `prompt-master-image-editor` (writes an English edit prompt only, never generates an actual image). • other prompt tasks → the relevant `prompt-master-*` skill. Also DO NOT TRIGGER for: Stable Diffusion / TensorArt generation (use `sd-image-gen`), Gemini / nano banana editing (use `nano-banana-pro-edit`), upscaling (use `sd-upscaler`), or image selection / judging (use `image-judge`).

中文简介正在准备中。当前显示原文说明。

Light-Novel-Style Title Architect (輕小說式書名企劃). Reads user material — proposal, logline, synopsis, script, character sheet or rough idea — and proposes 3–5 light-novel / web-novel-style work titles (書名・作品名・片名・劇名) for novels, web serials, anime, games, films and 微短劇, each with an abbreviation/hashtag, a cross-media short name and a check against spoilers, broken promises (タイトル詐欺) and copycat titles. Text only; never generates images. Use when the user says 「書名企劃」,「作品名發想」,「片名提案」,「劇名提案」,「書名發想」,「取書名」,「取片名」,「想書名」,「作品命名」,「輕小說書名」,「網文書名」,「長書名」,「タイトル案」, "title ideas", "name my novel", or asks what to call a story, script or project. Do NOT trigger for naming characters, places or factions (→ cultural-naming-director), social-post headlines (→ fb-post-architect), or code/file/skill naming.

中文简介正在准备中。当前显示原文说明。

Chief Movie Poster Art Director & Key Art Architect (電影海報主視覺總監). Takes ANY combination of user-supplied materials — a script / logline / plot synopsis, character images or character sheets, background & set images, stills, mood references — and designs THREE professional, visually arresting movie poster proposals that share ONE Key Art Bible (genre, 60-30-10 palette with hex codes, light logic, hero motif, locked character identity anchors, grade) yet use three DIFFERENT poster archetypes and three DIFFERENT composition archetypes — (1) 概念先導 Teaser, (2) 主視覺 Theatrical One-Sheet, (3) 角色／氛圍延伸 Character or Atmosphere poster. Each proposal ships a flowing natural-language English image-generation prompt (optimized for high-parameter NL models like Nano Banana / GPT-image2 / Seedream / Midjourney) for a TEXT-FREE key-art image with deliberately reserved negative space, a Traditional Chinese translation, and a separate typography & layout plan (片名處理、tagline、留白區、演職員字塊) — because titles and credits must be typeset over the image, never rendered by the model. Generates immediately, never asks questions. Use when the user mentions 「電影海報」,「海報設計」,「海報提示詞」,「海報企劃」,「宣傳海報」,「主視覺」,「電影主視覺」,「影展海報」,「角色海報」,「劇照海報」,「teaser 海報」,「key art」,「one-sheet」, "movie poster", "film poster", "poster design", "poster prompt", "key art", "theatrical one-sheet", "teaser poster", "character poster", or supplies a script plus character/background images and asks for poster concepts. It ONLY writes prompt and design-plan TEXT — it never generates images and never calls any generation API. Do NOT trigger for — 社群圖卡／資訊圖卡 (use social-media-visual-designer), 產品／商品圖 (use product-visual-architect), 分鏡 storyboard prompt sequences (use storyboard-director), 佈景／場景設計 (use set-design-director), 短影音腳本 (use prompt-master-short-video), or actual image generation with an explicitly named model such as gpt-image-2 / gemini-3-pro-image / nano banana pro (use those execution skills).

中文简介正在准备中。当前显示原文说明。

Anime-Realism Hybrid Lighting & Texture Director (High-Fidelity). Integrates anime aesthetics (2D character features with soft cel-shading) with hyper-realistic material rendering (PBR lighting/textures) for clothing, accessories, and backgrounds. Use this skill when the user mentions "高擬真", "材質渲染", "擬真材質", "寫實材質", "質感渲染", "hyper-realistic texture", "PBR", "hybrid rendering", "realistic textures on anime", or wants to combine anime character style with physically-based realistic material/lighting.

中文简介正在准备中。当前显示原文说明。

Advanced Natural Language Image Editing Architect — writes a high-density English IMAGE-EDITING PROMPT only; it NEVER generates or edits an actual image and NEVER calls any generation API or skill. THIS is the default skill to load for any generic image-editing request. Trigger when the user asks to edit an image or describes a hypothetical visual change, INCLUDING the Chinese phrases "編輯圖片", "圖片編輯", "載入圖片編輯", "圖片編輯技能", "改圖", "修圖", "編輯這張圖", "轉為黑白線稿", or English "edit image", "image editing", "modify the background" (e.g., "Change the car to red"). Do NOT trigger only when the user explicitly names an actual-generation model/provider such as gpt-image-2 / GMI / nano banana / gemini-3-pro-image / seedance — those run real generation, whereas this skill produces prompt text only.

中文简介正在准备中。当前显示原文说明。

Hailuo AI Visual Dynamics Architect (視覺動態架構師) for Image-to-Video (i2v) prompting. Takes a user-provided image (or image description) and applies cinematic physics + lens narrative to translate the static frame into highly-executable dynamic motion prompts. Always outputs TWO contrasting parallel proposals — Plan A (subtle / low-energy) and Plan B (kinetic / high-energy) — each with an English paragraph, a Traditional Chinese translation, and a structured summary (subject motion / background motion / camera movement). It ONLY writes motion-description TEXT; it never generates images, videos, or edits files. Use this skill when the user mentions "圖生影", "圖生視頻", "圖轉影片", "i2v", "img2video", "image2video", "image-to-video", "海螺圖生影", or wants to animate / add motion to a single still image into Hailuo-style (or similar i2v model) motion prompts. DO NOT trigger on generic "動態", "動態生成", "動態提示詞", "影片提示詞", or "text-to-video" — those belong to prompt-master-video-continuity; only trigger here when the request is explicitly image-to-video (i2v).

中文简介正在准备中。当前显示原文说明。

An AI Visual Alchemist & Scene Preservation Specialist skill that generates text prompts for transforming visual inputs into High-Contrast Lineless Watercolor style while maintaining environmental details. Use this skill when the user asks for "manga", "screentones", "漫畫", "網點", or "無線條水彩".

中文简介正在准备中。当前显示原文说明。

究極のオタクAI & Stable Diffusionプロンプト職人。ユーザーからの断片的・曖昧な指示（テキストや画像のアイデア）を基に、魅力的な「萌えキャラ」をデザインし、Stable Diffusion用の高品質な英語プロンプトを複数シーン分生成します。衣装デザイン時には自動的に costume-design-director スキルと連携します。

中文简介正在准备中。当前显示原文说明。

Engineers precision-grade, natural language text prompts (English + Traditional Chinese) for natural-language-aware image generation models. Strictly prohibits guessing or making up missing details. MUST use an 8-12 point deep-dive clarification process to extract specific aesthetic, physiognomy, and heritage details before synthesizing 3 distinct, production-grade prompt options. Use this skill when the user wants to create prompts for character design or photorealistic portraits. Keep skin descriptions natural, avoiding overly intricate pore/imperfection details.

中文简介正在准备中。当前显示原文说明。

Elite Animated Film Scene Architect & AI Prompt Engineer specializing in spatial reasoning and 360-degree environment extrapolation. Use this skill when the user provides a single 2D scene image and wants to generate a precise English YAML prompt for an AI image generation model (like Nano Banana, Seedream, Midjourney) to create a 2x2 4-panel grid showing four specific camera angles (front, left 90°, right 90°, back 180°) of the exact same scene while strictly maintaining the original art style, color palette, and lighting logic.

中文简介正在准备中。当前显示原文说明。

Material Reverse-Engineering & Texture Restoration Prompt Architect (材質逆向・紋理修復). Analyzes a degraded input image — noisy/messy background, smeared or over-retouched subject surface, plastic-looking skin, banding, compression damage — and writes ONE ready-to-run Traditional-Chinese Img2Img (圖生圖／全局重繪) redraw prompt that regenerates physically-real texture: natural elastic skin with fine pores & true individual hair strands for people; authentic brushed-metal / anodized / matte-plastic / glass / leather / wood for products; real leaf veins / concrete / fabric for scenes — while FORCING a clean, completely noise-free background and LOCKING the original lighting, perspective & composition for seamless fusion. Auto-appends the mandatory verbatim blocks for human subjects (局部重生強制指令) and products (產品精修). Use when the user mentions 「圖片紋理修復」,「肌膚紋理修復」,「紋理修復」,「材質修復」,「膚質修復」,「去髒」,「去噪」,「去油光」,「塑料感」,「重繪修復」,「圖生圖修復」,「Img2Img 修復」, "texture repair", "texture restoration", "skin texture restore", "material regeneration", "img2img cleanup / de-noise", or wants a degraded photo's surfaces/materials rebuilt into realistic texture via an image-to-image redraw prompt. It ONLY writes prompt TEXT — it NEVER generates or edits an actual image and NEVER calls any generation API/skill. Do NOT trigger when the user explicitly names an actual-generation model/provider (gpt-image-2 / GMI / gemini-3-pro-image / nano banana / seedance) — those run real generation; and prefer prompt-master-image-editor for generic non-texture edits (換背景／改顏色／加物件／轉線稿).

中文简介正在准备中。当前显示原文说明。

AI Video Continuity Director, Data Efficiency Architect & Vocal Sound Designer. Converts user concepts or reference images (character refs, set images, blocking/floor-plan diagrams) into camera-ready, audio-inclusive YAML prompts for video AI models — a 4-5 shot multi-shot sequence by default, or SINGLE-TAKE mode (一鏡到底) that renders ONE unbroken shot whose camera path is a time-coded beat list. Enforces strict visual continuity, character locking, telegraphic style, and a hard 3000-character YAML limit. Use this skill when the user mentions "動態", "動態生成", "動態提示詞", "影片提示詞", "video prompt", "動畫提示詞", "video continuity", "一鏡到底", "長鏡頭", "oner", "long take", "one-take", "single continuous take", or wants a compact single-block YAML prompt with audio tracks. This is the default skill for any "動態生成" request. DO NOT trigger for "文生影", "文生視頻", "t2v", "txt2video", "text2video", or requests for the paired single-take + multi-shot bilingual PROSE scripts — those belong to prompt-master-text-to-video.

中文简介正在准备中。当前显示原文说明。

Executable tool skill that ACTUALLY CALLS the GMI Cloud `seedance-2-0-260128` API to generate real videos — it runs a bundled Python script (submit → poll → download), it is NOT a prompt writer. Activation is by MODEL NAME ONLY. Trigger EXCLUSIVELY when the user explicitly names the model: "seedance 2", "seedance2", "seedance 2.0", "seedance-2-0", "seedance-2-0-260128", or "seedance 2.0 生成/出片". CRITICAL — DO NOT trigger on generic video-related wording: "影片生成", "文生影", "圖生影", "動態生成", "動態提示詞", "video generation", "text-to-video", "image-to-video", "i2v", "t2v", etc. — those describe VIDEO PROMPT tasks handled by prompt-master-text-to-video / prompt-master-image-to-video / prompt-master-video-continuity and are UNRELATED to this skill. Also do NOT trigger on bare "seedance" WITHOUT "2" (Seedance 1.x / generic prompt-target mentions belong to the prompt skills) nor on "seedance 2 fast" / "seedance 2.0 fast" (that is a different model, seedance-2-0-fast-260128). Only the explicit model name above activates this skill; if the user merely describes a video or asks for a prompt but never names this model, do NOT use this skill.

中文简介正在准备中。当前显示原文说明。

Chief Set Design Director & Production Designer (佈景指導) for narrative-driven scene/set prompts AND interior-design style prompts. Converts any space need — a script beat, a location idea, an era/genre/mood, a character's living or working space, or a named interior style (北歐風/日式/侘寂/Japandi/工業風/法式/Art Deco/飯店風/台式老宅… for living rooms, bedrooms, cafés, offices) — into THREE cohesive natural-language English image-generation prompts (optimized for high-parameter NL models like Nano Banana / GPT-image2 / Seedream) that share one Design Bible (era, style, class, 60-30-10 palette, materials, light logic, hero props) and cover the three professional distances — (1) establishing wide, (2) lived-in mid shot with props as silent dialogue, (3) texture & mood close-up — each paired with a Traditional Chinese translation. Generates immediately, never asks questions. Use when the user mentions 「佈景」,「佈景設定」,「佈景設計」,「場景設計」,「場景設定」,「場景提示詞」,「美術陳設」,「空間設計」,「室內設計」,「室內風格」,「風格空間」, a named interior style, "set design", "set dressing", "scene design", "scene prompt", "production design", "environment design", "interior design", "interior style", or asks what a character's room / home / office / hideout looks like. It ONLY writes prompt TEXT — it never generates images or calls any generation API. Do NOT trigger for: multi-view extrapolation of a provided 2D scene image into a 2x2 grid (use prompt-master-scene-architect), storyboard shot sequences 分鏡 (use storyboard-director), or actual image generation with an explicitly named model like gpt-image-2 / gemini-3-pro-image.

中文简介正在准备中。当前显示原文说明。

Elite Film Director & Screen-Translation Architect (分鏡導演／畫面轉譯). Takes a SHORT script / prose / dialogue snippet and translates it into a structured, shootable 分鏡腳本 (shot breakdown script) of 4–6 cuts (typically 4–5) in Traditional Chinese — each cut specifying 景別/攝影距離・角度 (shot size/distance & angle), 主體描述 (subject & action), 場景說明 (setting & props), 運鏡方式 (camera movement & cutting rhythm), 氛圍・情緒 (mood & subtext), and 台詞・演技 (dialogue in original language + acting beats). Grounded in prose→film adaptation craft — externalization/客觀對應物, subtext/潛台詞, information economy/懸念, time reconstruction, dialogue rewriting/台詞轉譯 (does it need saying? spoken-not-written syntax, voice differentiation, on-the-nose & "As you know Bob" avoidance, silence as its own line), and the "meaning lives in the gaps／意義活在縫隙裡" philosophy (reveal the least so the audience infers the rest). Use when the user mentions 「分鏡腳本」,「拍攝劇本轉譯」,「劇本轉分鏡」,「劇本分鏡」,「劇本轉譯」,「腳本轉分鏡」,「把劇本／台詞拆成鏡頭(cut)」, "shot script", "shot breakdown script", "script-to-shots", "screenplay breakdown", OR provides a short script/dialogue passage and wants it turned into a shot-by-shot cinematic script with camera language, mood, and acting. It ONLY writes the 分鏡腳本 TEXT — it never generates images/videos or calls any generation API. Do NOT trigger for: AI text-to-image storyboard PROMPTS (「分鏡設計」/「分鏡建議」/"storyboard" → use storyboard-director); 九宮格/nine-panel grid (→ prompt-master-9panel-grid); a deep script ANALYSIS report on structure/psychology/market (→ script-doctor-architect); video-generation prompts 動態/文生影/圖生影/t2v/i2v (→ prompt-master-text-to-video / image-to-video / video-continuity); or actual video generation with a named model like seedance (→ seedance-2-0).

中文简介正在准备中。当前显示原文说明。

Guide for creating effective skills. This skill should be used when users want to create a new skill (or update an existing skill) that extends Claude's capabilities with specialized knowledge, workflows, or tool integrations. You MUST ALWAYS load and trigger skill-optimizer and skill-translator to finalize the skill creation.

中文简介正在准备中。当前显示原文说明。

An advanced optimizing processor that executes after skill-creator or when a user wants to refactor an existing skill. It enforces Context Window Efficiency and the Progressive Disclosure pattern to ensure the resulting skill is lean, modular, and performant. Use this skill when requested to "optimize", "refactor", or "compress" an existing SKILL.md.

中文简介正在准备中。当前显示原文说明。

Generates human-readable Traditional Chinese (zh-TW) translations of skill files (SKILL.md and references). It operates on a progressive disclosure basis, storing the translations safely in a docs/ folder to avoid polluting the AI's context window. Use this skill immediately after skill-optimizer or whenever a user requests a translated/readable version of an existing skill.

中文简介正在准备中。当前显示原文说明。