Cover Tips

Cover Tips

Use this skill as a style-specific cover prompt organizer. It does not choose a style by guessing unless the user explicitly asks for a recommendation. The normal workflow is:

$cover-tips + style + --out-type template|prompt|all + user content

Default output is a standardized template. If --out-type is omitted, behave as --out-type template.

Supported Styles

Map user style names and aliases to these cover skills:

• Black-white minimal: cover-black-white-minimal
• Aliases: 黑白极简, 黑白, 极简, minimal, bw, black white, black-white.
• Use for rational, restrained, premium, monochrome, Swiss-grid, portfolio-grade covers.
• Trendy color poster: cover-trendy-color-poster
• Aliases: 潮流彩色, 彩色, 高冲击, trendy, color, color poster, 高饱和.
• Use for product covers, promotional posters, marketplace covers, launch visuals, bright high-impact graphics.
• Giant perspective poster: cover-giant-perspective-poster
• Aliases: 巨型透视, 透视标题, 透视海报, 电影海报风, 电竞主视觉, 运动品牌风, perspective, giant type, cinematic poster, 高冲突撞色, 中文透视.
• Use for giant Chinese perspective titles, cinematic/sports/esports key visuals, movie-poster composition, typography-as-architecture, and high-contrast clash covers where text is spatial structure.
• Sketch knowledge poster: cover-sketch-knowledge-poster
• Aliases: 手绘知识图谱, 手绘知识图谱风, 知识图谱, 知识地图, 白板框架, 手绘信息图, sketch knowledge, sketch knowledge poster, sketchnote, whiteboard framework.
• Use for hand-drawn knowledge maps, sketchnote infographic covers, whiteboard framework posters, product education covers, AI tool maps, capability maps, tutorial covers, and collectible educational diagram covers.
• 3D Eye: cover-3d-eye
• Aliases: 本地AI, 本地AI封面, 本地 AI, local AI, local ai terminal, terminal poster, 黑绿终端, 霓虹绿终端, Ollama教程, 离线AI, uncensored AI, private AI, offline AI.
• Use for dark local AI tutorial covers, terminal-native posters, offline/private/uncensored AI covers, Ollama/local model setup visuals, privacy-first AI covers, black-grid neon-green education posters, and mascot-led local ownership visuals.
• Midnight studio: cover-midnight-studio
• Aliases: 深夜工作室, 深夜工作室风, midnight studio, AI工程师空间, 独立开发者, indie hacker, 工作站, 多屏工作站, build in public, creator lab, coding at night.
• Use for cinematic late-night AI creator workspace covers with multi-monitor stations, premium tech photography, film lighting, and HUD-integrated titles—not RGB gaming rooms or ordinary desk photos.
• Light product: cover-light-product
• Aliases: 浅色产品, 浅色产品风, SaaS产品, light product, 奶油白, 产品风, 浅色Hero, 浅色产品美学.
• Use for AI product covers, SaaS brand headers, product launch visuals, Agent workspace covers, and light/cream-based premium product marketing visuals with warm-cool dual color fusion—not dark mode, not ordinary banner ads.
• McKinsey briefing style: cover-mckinsey-briefing-style
• Aliases: 麦肯锡简报风, 麦肯锡风, 咨询简报, 咨询简报封面, 战略简报, 战略报告, 咨询报告, consulting briefing, mckinsey style, strategy report, executive briefing, boardroom deck.
• Use for consulting-grade executive briefing covers, strategy report title pages, boardroom deck covers, business analysis visuals, and PPT consulting covers with strict grids, crisp white space, strategy frameworks, and disciplined blue-black-red accents.
• Pixel avatar: cover-pixel-avatar
• Aliases: 像素头像, 像素, 8-bit头像, 8bit头像, 8-bit avatar, pixel avatar, retro game avatar, NFT头像, Q版像素头像, 社交头像.
• Use for abstract retro 8-bit pixel art avatars redesigned from an uploaded reference image, with solid flat backgrounds, high-saturation clash colors, and commercial avatar/IP energy. Requires a reference image.
• Budapest poster: cover-budapest-poster
• Aliases: 布达佩斯, budapest, 复古欧洲, 复古中欧, 电影感, 明信片, 剧院感, 电车, 车站, 浴场.
• Use for retro Central European cinematic posters, theatre/tram/station/bathhouse/archive scenes, postcard-like concepts, and spatial storytelling covers. Hotel imagery is only a specific optional subset, not the default.
• Editorial collage: cover-editorial-collage
• Aliases: 撕纸剪贴, 剪贴, 拼贴, collage, editorial collage, torn paper, 杂志拼贴.
• Use for torn-paper collage, magazine collage, satire, social commentary, conflict, handmade editorial covers.
• Tea oriental: cover-tea-oriental
• Aliases: 茶风格, 茶, 东方美学, 宋代美学, 宋风, 文人气, 国风编辑设计, 汉字成像, 字中有画, tea, oriental tea, song literati.
• Use for high-end oriental cultural posters, tea-aesthetic invitations, character-as-image covers, refined infographics, PPT covers, and exhibition-style visuals with rice-paper texture and Song literati mood.

If no style is specified, ask the user to choose one of: 黑白极简, 潮流彩色, 巨型透视, 手绘知识图谱, 3D Eye, 深夜工作室, 像素头像, 布达佩斯, 撕纸剪贴, 茶风格, 浅色产品, 麦肯锡简报风. Do not silently pick a style unless the user asks for a recommendation.

cover-tips remains a cover-only router. If the user asks for a coordinated article cover plus inline visual package, direct them to article-visual-planner and preserve the selected cover-* style instead of trying to produce article visual briefs here.

Output Type

Use the explicit --out-type parameter to decide what to output. Read --out-type only as a control parameter, not as part of the user's topic or subtitle.

• --out-type template: output the standardized invocation template only.
• --out-type prompt: output the generic image prompt only.
• --out-type all: output the template first, then the generic image prompt.
• Omitted --out-type: default to template.

Backward compatibility:

• Treat 模版, 模板, 整理成格式, 标准格式, 调用格式, and 使用格式 as --out-type template.
• Treat explicit output requests such as 输出提示词, 生成提示词, 完整提示词, 生图提示词, 直接给 prompt, 只要提示词, 只要 prompt, and image prompt as --out-type prompt.
• Treat 模版和提示词, 模板和 prompt, 模版+提示词, 两个都要, 都输出, 先给模版再给提示词, and 既要标准格式也要完整 prompt as --out-type all.

Do not infer prompt mode from field values. For example, 主题：提示词 means the topic is "提示词"; it does not by itself mean --out-type prompt.

If --out-type has any value other than template, prompt, or all, ask the user to choose one of those three values.

Field Extraction

Extract or infer these fields from the user's content:

• 主题词: the core title. Preserve product names, regions, denominations, or exact conceptual titles.
• 副标题: a concise supporting line: benefits, promise, context, or secondary idea.
• 画幅比例: respect explicit user input. If omitted, infer from use case.
• 语言: Chinese, English, or mixed Chinese-English.
• 用途: X cover, marketplace cover, WeChat cover, poster, product poster, knowledge card, portfolio cover, etc.
• 情绪倾向: infer from content and selected style.
• 禁用元素: combine user-forbidden items with style defaults.

Ratio defaults:

• X cover: 5:2.
• Xianyu / marketplace cover: 5:2 unless the user explicitly asks for square marketplace format.
• WeChat cover: 3:2.
• Poster / event poster: 3:4.
• Portfolio cover: 4:5.
• Knowledge card: 1:1 or 4:5; use 1:1 when no platform is specified.

Language defaults:

• Mostly Chinese input: 中文.
• Mixed Chinese and English product/platform words: 中英混排.
• Mostly English input: 英文.

Style Defaults

Use these default moods and forbidden elements when the user does not specify them.

Black-white minimal:

• Mood: 理性 / 克制 / 可信 / 冷静 / 精密.
• Forbidden: 彩色 Apple logo、购物车图标、金币堆、蓝紫霓虹、机器人脸、廉价模板、过度 3D、大段说明文字.

Trendy color poster:

• Mood: 明亮 / 快速 / 可信 / 高冲击 / 活跃.
• Forbidden: 廉价电商模板、购物车图标、金币堆、蓝紫霓虹、机器人脸、低端 3D、PPT 风、过度堆字.

Giant perspective poster:

• Mood: 热血 / 速度 / 压迫 / 戏剧化 / 胜利 / 反叛.
• Forbidden: 小字标题、无透视平排标题、文字与插图分离、黑白灰平淡配色、廉价霓虹、蓝紫科技渐变、电商促销风、知识卡片风、PPT 封面、模板居中排版、中文错字.

Sketch knowledge poster:

• Mood: 高级 / 克制 / 专业 / 清晰 / 易收藏 / 教育感.
• Forbidden: 纯白背景、科技背景、深色背景、渐变背景、霓虹背景、赛博朋克背景、彩虹色、多彩配色、科技蓝、紫色、绿色、红色、金色、3D图标、CAD风格、企业图标库风格、PPT模板、SaaS Banner、信息堆砌、不可读小字.
• Ratio defaults: X 5:2, WeChat header 2.35:1, tutorial/blog 16:9, knowledge card 1:1, poster 3:4.

Local AI terminal:

• Mood: 私密 / 本地掌控 / 终端感 / 轻微反叛 / 教程清晰 / 黑客式克制.
• Forbidden: 紫蓝霓虹、赛博朋克城市、Matrix代码雨、RGB电竞房、黑客兜帽人、机器人脸、廉价3D图标、杂乱仪表盘、官方产品logo、不可读小字、逐字复制案例图文字.
• Ratio defaults: X 5:2, WeChat header 2.35:1, tutorial/blog/YouTube 16:9, knowledge card 1:1, poster 3:4.

Midnight studio:

• Mood: 专注 / 孤独 / 高级感 / 深夜平静 / 实验感 / 创业拼搏 (infer from topic when possible).
• Forbidden: RGB电竞房、花哨灯带、低级霓虹、廉价赛博朋克、电商摄影感、样板间、网吧、电竞酒店、宿舍、普通电脑桌、标题贴纸感、杂乱堆屏.
• Ratio defaults: X 5:2, Xiaohongshu 3:4, WeChat 2.35:1, Xianyu 1:1, Bilibili 16:9, blog 21:9.

Budapest poster:

• Mood: 复古 / 戏剧感 / 精致 / 轻微荒诞 / 电影感.
• Forbidden: 机器人脸、蓝紫霓虹、廉价科技感、普通旅行明信片、官方品牌标志、过度 3D、杂乱拼贴.

Editorial collage:

• Mood: 讽刺 / 冲突 / 街头 / 复古 / 观点感.
• Forbidden: 机器人脸、蓝紫霓虹、廉价科技感、PPT 布局、干净矩形堆叠、低质脏乱朋克、不可读文字.

Tea oriental:

• Mood: 安静 / 雅致 / 书卷气 / 东方哲思 / 展览感.
• Forbidden: 廉价国风模板、俗气红金配色、随机书法背景、过度装饰边框、网红茶室风、杂乱信息、不可读核心字、蓝紫霓虹、低端 3D.

Light product:

• Mood: 专业 / 高级感 / 未来工作方式 / SaaS品质 / 温暖但精密.
• Forbidden: 暗黑模式、咖啡杯、笔记本、铅笔、人手打字、芯片电路板、机器人、发光大脑、赛博朋克、宇宙星空、廉价AI海报风、电商封面感、荧光渐变色、只有暖色或只有冷色.
• Ratio defaults: X 5:2, Xiaohongshu 3:4, WeChat 2.35:1, Xianyu 1:1, Bilibili 16:10, blog 16:9.

McKinsey briefing style:

• Mood: 理性 / 权威 / 精准 / 克制 / 战略感 / 董事会级.
• Forbidden: 官方麦肯锡logo、官方品牌标识、虚假隶属关系、握手图、商务人士摆拍、摩天楼图库、蓝紫霓虹、机器人脸、金币堆、廉价PPT模板、拥挤信息图、随机漂浮图表、过度3D.
• Ratio defaults: X/LinkedIn 5:2, PPT 16:9, WeChat header 2.35:1, report title page 4:5, knowledge card 1:1.

Pixel avatar:

• Mood: 复古 / 抽象 / 俏皮 / 高饱和 / 游戏感 / 社交头像感.
• Forbidden: 写实照片、3D、厚涂、普通动漫插画、真实照片质感、复杂背景、保留原图场景、风景、街道、建筑、房间、桌面、草地、天空、过度追求相似度、固定黄底、固定橙发、低对比脏色.
• Fixed ratio: 1:1; requires an uploaded reference image.

Template Output

For template-only mode, output exactly this structure and fill every field:

使用 $cover-{selected-style} 生成一张封面
主题词：{topic}
副标题：{subtitle}
画幅比例：{ratio}
语言：{language}
用途：{use_case}
情绪倾向：{mood}
禁用元素：{forbidden_elements}

Skill invocation prefix

Detect the current platform and use the matching prefix in the template:

• Claude Code → /cover-{style}
• Codex → $cover-{style}
• Cursor → /cover-{style}
• Gemini CLI → /cover-{style}
• Default (unknown platform) → /cover-{style}

Generic Prompt Output

For prompt-only mode, output a generic image-generation prompt. Do not include provider-specific syntax, model names, API names, or agent-specific parameters. Do not mention OpenAI, Claude, Midjourney, Flux, Stable Diffusion, or ComfyUI unless the user explicitly asks.

Build the prompt in the style of the selected cover skill, using its visual language and aspect-ratio discipline. The prompt should include:

1. Exact canvas ratio and canvas size in the first line.
2. Topic, use case, language, A/B/C title layers.
3. Style description.
4. Visual metaphor and 1-3 concrete anchors.
5. Text-image relationship.
6. Typography hierarchy.
7. Color/background system.
8. Mood and extra context.
9. Avoid list.

Do not hard-code one provider's prompt syntax. Use plain English image-prompt language that any image-generation agent can adapt.

Both Output

When both template and prompt are requested, output:

【模版】
{template}

【提示词】
{generic image prompt}

Error Handling

• If style is missing: ask the user to choose one style and list the supported styles: 黑白极简, 潮流彩色, 巨型透视, 手绘知识图谱, 3D Eye, 深夜工作室, 像素头像, 布达佩斯, 撕纸剪贴, 茶风格, 浅色产品, 麦肯锡简报风.
• If user content is missing: ask for at least a title, topic, product description, or article idea.
• If --out-type is missing or unclear: default to template.
• If the user asks for automatic style recommendation, provide one recommended style and then output the requested mode.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: imartinstudio
• Source: imartinstudio/cover-prompt-skills
• License: MIT

Install and usage instructions live in the source repository linked above.