← 返回 Skills 市场
1104
总下载
2
收藏
6
当前安装
1
版本数
在 OpenClaw 中安装
/install logo
功能描述
Generate logos with AI image tools using effective prompt structures, validation loops, and export workflows for App Store icons and brand marks.
使用说明 (SKILL.md)
Quick Start: AI Logo Generation
Best model for most logos: Nano Banana Pro (Gemini 3 Pro Image)
Basic Prompt Formula
Create a [STYLE] logo featuring [ELEMENT] on [BACKGROUND].
[DESCRIPTION]. The logo should look good at 32px with recognizable shapes.
Example
Create a minimalist logo featuring a geometric mountain peak on white background.
Clean lines, navy blue (#1E3A5A), modern and professional style.
The logo should look good at 32px with recognizable shapes.
For the full 7-step prompt framework and model comparison, load ai-generation.md.
Decision Tree
| Situation | Load |
|---|---|
| AI generation (Nano Banana, GPT Image, prompts, iOS icons) | ai-generation.md |
| Logo types (wordmark, symbol, combo, emblem) | types.md |
| Design process with a human designer | process.md |
| File formats and export requirements | formats.md |
| DIY without AI (templates, Canva) | diy.md |
| Hiring designers or agencies | hiring.md |
Model Quick Reference
| Model | Best For |
|---|---|
| Nano Banana Pro | Overall best, text + icons, App Store icons |
| GPT Image 1.5 | Conversational iteration, natural language |
| Ideogram | Perfect text rendering |
| Midjourney v7 | Artistic icons only (no text) |
iOS App Icons (Liquid Glass)
iOS 26 uses Liquid Glass design. Use this prompt structure:
Create a polished iOS app icon featuring [ELEMENT].
Rounded square with [COLOR] gradient, minimalist white symbol centered.
Soft shadows, glassy depth effect, works at 60px.
The icon represents [APP PURPOSE].
See ai-generation.md for the complete iOS 26 prompt template.
Validation Loop (MANDATORY)
NEVER deliver without visual review. Every AI output must be inspected before sharing.
- Generate → 2. Look at the actual image → 3. Check for issues → 4. Fix or regenerate → 5. Repeat (max 5-7 attempts)
Common fixes:
- Unwanted padding → Crop
- Elements cut off → Regenerate with "centered composition"
- Text garbled → Use Nano Banana/Ideogram or add manually
- Too complex → Simplify prompt
If 5-7 attempts fail, change model or strategy entirely.
Universal Truths
AI output is a starting point. Every AI logo needs vectorization, cleanup, and manual text refinement. Never use raw output as final.
Test at small sizes early. If it doesn't work at 32px, simplify. Most real-world usage is small.
Text handling varies. Only Nano Banana and Ideogram reliably render text. For Midjourney, generate icon-only.
Simple logos last. Nike, Apple, McDonald's. Complexity dates quickly and fails at small sizes.
Before Finalizing
- Works in black and white
- Readable at 32px (favicon test)
- Vectorized (SVG), not just PNG
- All variants created (horizontal, stacked, icon-only)
- Text manually refined, not AI-generated
- Tested on dark and light backgrounds
When to Load More
| Situation | Reference |
|---|---|
| Full prompt frameworks, model comparison, iOS icons | ai-generation.md |
| Wordmark vs symbol vs emblem decisions | types.md |
| Working with designers, brief templates | process.md |
| SVG, PNG, favicon, size requirements | formats.md |
| Track what works, learn from iterations | feedback.md |
安全使用建议
This skill is a how-to guide for generating and finalizing logos with AI—its instructions are coherent and focused. Before installing or using it: (1) confirm you have legitimate access to whatever image/model services you plan to use (the skill mentions commercial models but doesn't provide API keys); (2) follow the mandatory visual-review and vectorization steps—do not use raw AI output as a final logo; (3) check licensing and ownership terms for any AI-generated assets and for third‑party tools like vectorizer.ai; and (4) be mindful that the agent could call external image/model services when invoked (normal behavior for an instruction skill)—ensure that fits your policies and that you consent to any external API usage.
功能分析
Type: OpenClaw Skill
Name: logo
Version: 1.0.0
The OpenClaw AgentSkills skill bundle for 'logo' is benign. All files (`_meta.json`, `SKILL.md`, `ai-generation.md`, `diy.md`, `feedback.md`, `formats.md`, `hiring.md`, `process.md`, `types.md`) contain documentation and instructions solely focused on AI-driven logo generation, design principles, workflow, and quality control. There are no executable scripts, suspicious network calls, file system access instructions beyond conceptual organization, or prompt injection attempts designed to subvert the AI agent's purpose or perform malicious actions. The content consistently aligns with the stated goal of generating and refining logos using AI tools.
能力评估
Purpose & Capability
Name/description match the contents: the docs provide prompt frameworks, model recommendations, validation loops, export workflows and design process guidance. Nothing in the files asks for unrelated access (no cloud creds, no system files, no unrelated binaries).
Instruction Scope
Runtime instructions stay within logo generation and delivery: prompt templates, mandatory visual review, iteration limits, vectorization and export steps. The skill references external models and web tools (e.g., Gemini/Nano Banana Pro, GPT Image, Ideogram, Midjourney, vectorizer.ai) but does not instruct reading system files or exfiltrating data. Note: it assumes access to these external image/model services but does not declare credentials.
Install Mechanism
No install spec and no code files—lowest risk. Nothing will be downloaded or written to disk by the skill itself.
Credentials
Requires no environment variables, credentials, or config paths. The only external dependencies are model/service usage and third-party web tools mentioned in the guidance; those are reasonable for an image-generation workflow but the skill does not provide or require API keys itself.
Persistence & Privilege
always:false and no install behavior. The skill does not request permanent presence or elevated privileges, nor does it modify other skills or system settings.
如何使用
- 确保已安装 OpenClaw（本地或 Docker 部署）
- 在对话框中输入安装命令：
/install logo - 安装完成后，直接呼叫该 Skill 的名称或使用
/logo触发 - 根据 Skill 的参数说明提供必要输入，即可获得结构化输出
版本历史
v1.0.0
Initial release
元数据
常见问题
Logo 是什么？
Generate logos with AI image tools using effective prompt structures, validation loops, and export workflows for App Store icons and brand marks. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 1104 次。
如何安装 Logo？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install logo」即可一键安装，无需额外配置。
Logo 是免费的吗？
是的，Logo 完全免费（开源免费），可自由下载、安装和使用。
Logo 支持哪些平台？
Logo 跨平台运行，可在任意部署了 OpenClaw / Claude Code 的环境中使用（cross-platform）。
谁开发了 Logo？
由 Iván（@ivangdavila）开发并维护，当前版本 v1.0.0。
推荐 Skills