npx skills add https://github.com/bbusenius/pygmalion --skill logo-design此技能指导创建独特、专业的标志,避免通用美学。所有标志均以纯 SVG 代码形式创建,不使用 AI 图像生成。每个标志交付都包含强制性的颜色变体和用于客户审阅的展示页面。
用户提供标志要求:品牌名称、用途、行业、基调或其他关于其需求的背景信息。
在创建之前,理解背景并确定一个清晰的方向:
关键:选择一个清晰的概念方向并精确执行。标志必须感觉是为这个特定品牌有意设计的,而不是像一个通用模板。
根据最适合品牌需求的情况,支持所有标志类型:
选择最适合品牌需求、使用场景和名称特征的标志类型。
在选择字体之前:使用 Skill 工具并设置 skill="frontend-design" 来检查 frontend-design 技能是否存在
绝对不要在标志中使用这些字体 - 它们通用、过度使用且不专业:
This skill guides creation of distinctive, professional logos that avoid generic aesthetics. All logos are created as pure SVG code - no AI image generation. Every logo delivery includes mandatory color variations and a presentation page for client review.
The user provides logo requirements: brand name, purpose, industry, tone, or other context about what they need.
Before creating, understand the context and commit to a clear direction:
CRITICAL : Choose a clear conceptual direction and execute it with precision. The logo must feel intentionally designed for this specific brand, not like a generic template.
Support all logo types based on what best serves the brand:
Choose the type that best fits the brand's needs, usage contexts, and name characteristics.
广告位招租
在这里展示您的产品或服务
触达数万 AI 开发者,精准高效
使用任何这些字体都是严重失误。 使用这些字体的标志看起来业余且容易被遗忘。
在选择字体之前,检查系统上可用的字体:
fc-list : family | sort -u
从结果中挑选独特的字体。Linux 上常见的好选项:
特别针对标志排版:
fc-list 检查可用性对于文字标志和字母标志,排版就是标志本身。每个曲线、末端和比例都必须是有意为之。
任何标志设计最多使用 3 种颜色。标志必须能用更少的颜色工作,而不是更多。
每个标志必须交付三个颜色版本:
所有三个版本必须使用相同的设计 - 仅改变颜色。单色版本证明标志在不依赖颜色传达意义的情况下也能正常工作。
在适当的时候,提供多种布局版本:
并非每个标志都需要所有变体。在以下情况下提供它们:
所有标志必须创建为干净的 SVG 代码:
<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 [width] [height]">
<!-- Logo content -->
</svg>
viewBox 实现可伸缩性,而不是固定宽度/高度<g> 标签对相关元素进行分组id="primary-color", id="accent-color"每个标志必须在以下尺寸下正常工作:
关键:保持宽高比
标志和图标在调整大小时绝不能被扭曲。所有调整大小操作必须保持原始比例:
preserveAspectRatio生成图形元素的两种方法:
gemini_generate_svg):直接生成 SVG 代码trace_bitmap):将栅格图像转换为 SVG 矢量图根据设计要求和可用资源进行选择。
当 gemini_generate_svg 工具可用时,它可以根据文本描述直接生成 SVG 代码。
使用gemini_generate_svg 用于:
关键:始终向 Gemini 请求单色(单一颜色)SVG。 该工具生成一个单色图形元素。然后您需要手动创建三个颜色变体。
Gemini 提示词示例:
"A minimalist mountain icon with three peaks, clean geometric shapes,
single color black (#000000), suitable for a coffee company logo"
当 trace_bitmap 工具可用时,它可以将栅格图像(PNG, JPG, BMP)转换为干净的 SVG 矢量图。这模拟了设计师描摹位图参考的传统工作流程。
使用trace_bitmap 用于:
位图描摹工作流程:
gemini_generate_image 创建位图 → trace_bitmap 进行矢量化turdsize:抑制斑点(默认值:2,值越大越干净)turnpolicy:路径提取(默认值:minority)alphamax:角点锐度(默认值:1.0,值越低越锐利)opttolerance:曲线平滑度(默认值:0.2,值越高越平滑)位图→描摹工作流程示例:
User: Design a logo for Mountain Coffee Co. using this mountain photo I have.
1. Use trace_bitmap on user's mountain.jpg → get mountain-traced.svg (black paths)
2. Clean/simplify the traced SVG if needed
3. Create three color versions:
- monotone: black mountain
- two-color: navy mountain with brown accent
- three-color: navy mountain, brown accent, cream highlight
4. Code the "Mountain Coffee Co." wordmark
5. Compose each color variation with the wordmark
6. Generate horizontal, vertical, and abbreviated layouts
7. Build presentation HTML showing all variations
生成→描摹工作流程示例:
User: Create a logo with an organic leaf illustration.
1. Use gemini_generate_image: "detailed organic leaf illustration, single leaf,
black and white, high contrast, clean edges, 1024x1024px"
2. Use trace_bitmap on the generated image → get leaf-traced.svg
3. Refine the traced paths if needed
4. Create three color versions (monotone, two-color, three-color)
5. Compose with brand typography
6. Build complete presentation
关键:集成 Potrace 输出
使用 trace_bitmap 描摹后(该工具包含自动的 viewBox 优化),您将得到一个类似这样的 SVG:
<?xml version="1.0" encoding="UTF-8"?>
<svg viewBox="0 0 1291.5 724" width="1291.5pt" height="724pt" ...>
<g transform="matrix(0.1,0,0,-0.1,-378.97,1370)" fill="#000000">
<path d="m 10830,13684 c -158,-41..."/>
</g>
</svg>
描摹后的 SVG 具有 viewBox 尺寸(例如,1291.5 x 724)和 <g> 元素上的矩阵变换。
要使用<g> 元素将此图标组合到标志中(允许 Inkscape 优化):
<svg xmlns="http://www.w3.org/2000/svg">
<!-- Traced icon wrapped in positioning transform -->
<!-- Calculate scale: desired_size / viewBox_dimension -->
<!-- Example: want 120px wide, viewBox is 1291.5, scale = 120/1291.5 = 0.0929 -->
<g transform="translate(20, 40) scale(0.0929)">
<!-- Copy the <g> and <path> EXACTLY from traced SVG -->
<g transform="matrix(0.1,0,0,-0.1,-378.97,1370)" fill="#1a1a1a">
<path d="m 10830,13684 c -158,-41..."/>
</g>
</g>
<!-- Typography -->
<text x="160" y="80" font-size="32">VOLCANO</text>
<text x="160" y="115" font-size="16">ADVENTURES</text>
</svg>
关键计算步骤:
translate(x_pos, y_pos) scale(calculated_scale)<g transform="matrix(...)"> 和 <path> - 不要修改它们创建标志后,使用 Inkscape 优化 viewBox:
inkscape logo.svg --export-area-drawing --export-plain-svg --export-filename=logo.svg
这种方法:
<g> 元素(没有嵌套的 <svg>)绝对不要做这些:
<path> 元素 - 您会丢失坐标系<svg> - 这会破坏 Inkscape 的边界框检测直接生成 SVG 代码(不使用 AI 工具)用于:
两种方法都只生成一个图形元素 - 而不是完整的标志或其变体。您始终负责:
绝不创建以下标志:
避免常见的 AI 标志美学:
阶段 1:设置和排版
阶段 2:创建可编辑版本 6. 创建基础 SVG 设计:不使用 viewBox,使用独特的字体(不使用禁止的字体) 7. 生成所有变体:单色、双色、三色、水平、垂直、仅图标 8. 将所有文件保存为可编辑版本:所有文件以 _editable.svg 后缀保存,文本保持为文本
阶段 3:转换为交付版本 9. 对于每个可编辑文件:
_editable 后缀)阶段 4:展示 10. 构建展示页面:HTML 页面显示最终交付版本 11. 在多种尺寸下测试:验证从网站图标到大尺寸格式的可读性
关键:您必须创建两个版本:
*_editable.svg):文本保持为文本,用于未来修改*.svg):文本已转换为轮廓,在任何地方都能完美渲染如果您跳过了文本到轮廓的转换,标志在浏览器中会出现渲染问题。
关键:在交付之前将所有文本转换为路径,以确保在浏览器和平台之间的一致渲染。
步骤 1:创建包含文本的可编辑版本
# Write the SVG with text elements (no viewBox)
# Save as: logo_monotone_horizontal_editable.svg
步骤 2:将文本转换为轮廓
inkscape logo_monotone_horizontal_editable.svg \
--actions="select-all:text;object-to-path" \
--export-plain-svg \
--export-filename=logo_monotone_horizontal_temp.svg
步骤 3:优化 viewBox
inkscape logo_monotone_horizontal_temp.svg \
--export-area-drawing \
--export-plain-svg \
--export-filename=logo_monotone_horizontal.svg
用于展示和使用:
用于客户未来编辑:
*_editable.svg)关键: 当使用描摹的位图创建组合标志时,使用 <g> 元素(而不是嵌套的 <svg>),以便 Inkscape 能够正确计算边界。
将文本转换为轮廓后,使用 Inkscape 优化 viewBox:
inkscape input.svg --export-area-drawing --export-plain-svg --export-filename=output.svg
此命令:
将此应用于每个标志文件(单色、双色、三色、水平、垂直、缩写),在将文本转换为轮廓之后。
在 HTML 展示页面中交付标志,展示以下内容:
并排显示所有三个颜色版本:
一起显示水平、垂直和缩写版本
在多种尺寸下显示标志:
<!DOCTYPE html>
<html>
<head>
<title>[Brand] Logo Presentation</title>
<style>
/* Grid layout for variations */
/* Light and dark background sections */
/* Size comparison displays */
</style>
</head>
<body>
<h1>[Brand] Logo</h1>
<section class="color-variations">
<h2>Color Versions</h2>
<!-- Monotone, Two-color, Three-color side by side -->
</section>
<section class="layout-variations">
<h2>Layout Options</h2>
<!-- Horizontal, Vertical, Abbreviated -->
</section>
<section class="size-tests">
<h2>Size Testing</h2>
<!-- Multiple sizes from small to large -->
</section>
<section class="background-tests">
<h2>Background Compatibility</h2>
<!-- Logo on light and dark backgrounds -->
</section>
</body>
</html>
如果安装了 frontend-design 技能,请将此技能与它一起调用,以获得全面的美学指导。frontend-design 技能提供以下方面的指导:
同时使用这两个技能以获得最大的设计质量。
每周安装次数
1
仓库
首次出现
1 天前
安全审计
安装于
amp1
cline1
opencode1
cursor1
kimi-cli1
codex1
BEFORE SELECTING FONTS: Check if the frontend-design skill exists using the Skill tool with skill="frontend-design"
ABSOLUTELY NEVER use these fonts in logos - they are generic, overused, and unprofessional:
Using any of these fonts is a critical failure. Logos with these fonts look amateurish and forgettable.
Before selecting fonts, check what's available on the system:
fc-list : family | sort -u
Pick distinctive fonts from the results. Common good options on Linux:
For logo typography specifically:
fc-list to check availabilityFor wordmarks and lettermarks , typography IS the logo. Every curve, terminal, and proportion must be intentional.
Maximum 3 colors in any logo design. Logos must work with fewer colors, not more.
Every logo must be delivered in three color versions:
All three versions must use the SAME design - only colors change. The monotone version proves the logo works without relying on color for meaning.
When appropriate, provide multiple layout versions:
Not every logo needs all variations. Provide them when:
All logos must be created as clean SVG code:
<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 [width] [height]">
<!-- Logo content -->
</svg>
viewBox for scalability, not fixed width/height<g> tagsid="primary-color", id="accent-color"Every logo must work at:
CRITICAL: Maintain Aspect Ratio
Logos and icons must NEVER be distorted when resized. All resizing must maintain the original proportions:
preserveAspectRatio when embeddingTwo approaches for generating graphical elements:
gemini_generate_svg): Direct SVG code generationtrace_bitmap): Convert raster images to SVG vectorsChoose based on the design requirements and available resources.
When the gemini_generate_svg tool is available , it can generate SVG code directly from text descriptions.
Usegemini_generate_svg for:
CRITICAL: Always request monotone (single color) SVGs from Gemini. The tool produces ONE graphical element in a single color. You then manually create the three color variations.
Example Gemini prompt:
"A minimalist mountain icon with three peaks, clean geometric shapes,
single color black (#000000), suitable for a coffee company logo"
When the trace_bitmap tool is available , it can convert raster images (PNG, JPG, BMP) to clean SVG vectors. This mirrors the traditional designer workflow of tracing bitmap references.
Usetrace_bitmap for:
Workflow for bitmap tracing:
gemini_generate_image to create bitmap → trace_bitmap to vectorizeturdsize: Suppress speckles (default: 2, higher = cleaner)turnpolicy: Path extraction (default: minority)alphamax: Corner sharpness (default: 1.0, lower = sharper)opttolerance: Curve smoothness (default: 0.2, higher = smoother)Example bitmap→trace workflow:
User: Design a logo for Mountain Coffee Co. using this mountain photo I have.
1. Use trace_bitmap on user's mountain.jpg → get mountain-traced.svg (black paths)
2. Clean/simplify the traced SVG if needed
3. Create three color versions:
- monotone: black mountain
- two-color: navy mountain with brown accent
- three-color: navy mountain, brown accent, cream highlight
4. Code the "Mountain Coffee Co." wordmark
5. Compose each color variation with the wordmark
6. Generate horizontal, vertical, and abbreviated layouts
7. Build presentation HTML showing all variations
Example generate→trace workflow:
User: Create a logo with an organic leaf illustration.
1. Use gemini_generate_image: "detailed organic leaf illustration, single leaf,
black and white, high contrast, clean edges, 1024x1024px"
2. Use trace_bitmap on the generated image → get leaf-traced.svg
3. Refine the traced paths if needed
4. Create three color versions (monotone, two-color, three-color)
5. Compose with brand typography
6. Build complete presentation
CRITICAL: Integrating Potrace Output
After tracing with trace_bitmap (which includes automatic viewBox optimization), you'll have an SVG like this:
<?xml version="1.0" encoding="UTF-8"?>
<svg viewBox="0 0 1291.5 724" width="1291.5pt" height="724pt" ...>
<g transform="matrix(0.1,0,0,-0.1,-378.97,1370)" fill="#000000">
<path d="m 10830,13684 c -158,-41..."/>
</g>
</svg>
The traced SVG has viewBox dimensions (e.g., 1291.5 x 724) and a matrix transform on the <g> element.
To compose this into a logo using<g> elements (allows Inkscape optimization):
<svg xmlns="http://www.w3.org/2000/svg">
<!-- Traced icon wrapped in positioning transform -->
<!-- Calculate scale: desired_size / viewBox_dimension -->
<!-- Example: want 120px wide, viewBox is 1291.5, scale = 120/1291.5 = 0.0929 -->
<g transform="translate(20, 40) scale(0.0929)">
<!-- Copy the <g> and <path> EXACTLY from traced SVG -->
<g transform="matrix(0.1,0,0,-0.1,-378.97,1370)" fill="#1a1a1a">
<path d="m 10830,13684 c -158,-41..."/>
</g>
</g>
<!-- Typography -->
<text x="160" y="80" font-size="32">VOLCANO</text>
<text x="160" y="115" font-size="16">ADVENTURES</text>
</svg>
Key calculation steps:
translate(x_pos, y_pos) scale(calculated_scale)<g transform="matrix(...)"> and <path> exactly - don't modify themAfter creating the logo, optimize viewBox with Inkscape:
inkscape logo.svg --export-area-drawing --export-plain-svg --export-filename=logo.svg
This approach:
<g> elements throughout (no nested <svg>)NEVER do these:
<path> element - you lose the coordinate system<svg> - it breaks Inkscape's bounding box detectionGenerate SVG code directly (without AI tools) for:
Both approaches produce a single graphical element - not the complete logo or variations. You are always responsible for:
NEVER create logos that:
AVOID common AI logo aesthetics:
Phase 1: Setup and Typography
Phase 2: Create Editable Versions 6. Create base SVG design : WITHOUT viewBox, using distinctive fonts (NO forbidden fonts) 7. Generate all variations : Monotone, two-color, three-color, horizontal, vertical, icon-only 8. Save all as editable : All files saved with _editable.svg suffix, text as text
Phase 3: Convert to Delivery Versions 9. For EACH editable file :
_editable suffix)Phase 4: Presentation 10. Build presentation page : HTML showing final delivery versions 11. Test at multiple sizes : Verify legibility from favicon to large format
CRITICAL: You must create BOTH versions:
*_editable.svg): Text as text, for future modifications*.svg): Text converted to outlines, perfect rendering everywhereIf you skip the text-to-outlines conversion, logos will have rendering issues in browsers.
CRITICAL : Convert all text to paths before delivery to ensure consistent rendering across browsers and platforms.
Step 1: Create editable version with text
# Write the SVG with text elements (no viewBox)
# Save as: logo_monotone_horizontal_editable.svg
Step 2: Convert text to outlines
inkscape logo_monotone_horizontal_editable.svg \
--actions="select-all:text;object-to-path" \
--export-plain-svg \
--export-filename=logo_monotone_horizontal_temp.svg
Step 3: Optimize viewBox
inkscape logo_monotone_horizontal_temp.svg \
--export-area-drawing \
--export-plain-svg \
--export-filename=logo_monotone_horizontal.svg
For presentation and use:
For client's future edits:
*_editable.svg)CRITICAL: When creating composed logos with traced bitmaps, use <g> elements (NOT nested <svg>) so Inkscape can calculate bounds correctly.
After converting text to outlines, optimize the viewBox using Inkscape:
inkscape input.svg --export-area-drawing --export-plain-svg --export-filename=output.svg
This command:
Apply this to EVERY logo file (monotone, two-color, three-color, horizontal, vertical, abbreviated) AFTER converting text to outlines.
Deliver logos in an HTML presentation page that shows:
Display all three color versions side-by-side:
Display horizontal, vertical, and abbreviated versions together
Show the logo at multiple sizes:
<!DOCTYPE html>
<html>
<head>
<title>[Brand] Logo Presentation</title>
<style>
/* Grid layout for variations */
/* Light and dark background sections */
/* Size comparison displays */
</style>
</head>
<body>
<h1>[Brand] Logo</h1>
<section class="color-variations">
<h2>Color Versions</h2>
<!-- Monotone, Two-color, Three-color side by side -->
</section>
<section class="layout-variations">
<h2>Layout Options</h2>
<!-- Horizontal, Vertical, Abbreviated -->
</section>
<section class="size-tests">
<h2>Size Testing</h2>
<!-- Multiple sizes from small to large -->
</section>
<section class="background-tests">
<h2>Background Compatibility</h2>
<!-- Logo on light and dark backgrounds -->
</section>
</body>
</html>
If the frontend-design skill is installed , invoke it alongside this skill for comprehensive aesthetic guidance. The frontend-design skill provides direction on:
Use both skills together for maximum design quality.
Weekly Installs
1
Repository
First Seen
1 day ago
Security Audits
Installed on
amp1
cline1
opencode1
cursor1
kimi-cli1
codex1
前端设计技能指南:避免AI垃圾美学,打造独特生产级界面
39,800 周安装