← 返回 Skills 市场
325
总下载
0
收藏
0
当前安装
1
版本数
在 OpenClaw 中安装
/install limtdesign
功能描述
视觉创意生图提示词生成skill。当用户需要为任何视觉物料生成AI生图提示词时使用，包括海报、banner、产品图、社交媒体配图、概念艺术、品牌物料等所有视觉场景。无论用户的需求是模糊的（"帮我做张海报"）、半清晰的（"科技风新品发布海报"）还是已有方向的，都应触发此skill。当用户提到生图、出图、画图、提示词...
使用说明 (SKILL.md)
视觉创意生图 Skill 主文件
概述
本skill帮助AI将用户的视觉需求转化为富有创意的AI生图提示词。核心目标是突破平庸，解决三个常见问题：
- 创意局限：避免走最安全、最常见的视觉解法
- 构图死板：避免均匀分布和居中对称的惰性构图
- 画面空洞：主动填充丰富的视觉细节和层次
处理流程：分析需求 → 识别场景 → 调取创意方法 → 生成设计思路（用户可见）→ 输出提示词（传递下游）
如当前需求属于特定场景，请同时读取对应的场景子文件以获取场景专属方法：
- 场景子文件位于
scenes/目录下 - 见本文件末尾的场景索引
八、场景子文件索引
识别用户需求后，读取对应场景子文件获取场景专属创意方向：
| 场景 | 子文件路径 |
|---|---|
| 电商主图与详情页 | scenes/ecommerce.md |
| 促销海报与活动banner | scenes/promotion.md |
| 户外广告OOH | scenes/ooh.md |
| 产品概念图与产品海报 | scenes/product.md |
| 品牌视觉与品牌故事图 | scenes/brand.md |
| 包装设计概念 | scenes/packaging.md |
| 通用海报设计 | scenes/general-poster.md |
| 小红书视觉（种草/生活/教程/探店） | scenes/xiaohongshu.md |
| 抖音视觉（封面/直播间/主页） | scenes/douyin.md |
| 微信公众号视觉（封面/配图/海报） | scenes/wechat-oa.md |
| 微博视觉（营销/热点/长图/九宫格） | scenes/weibo.md |
| 短视频封面与缩略图 | scenes/video-cover.md |
| 信息图与数据可视化 | scenes/infographic.md |
| 电影与剧集海报 | scenes/film-poster.md |
| 游戏视觉与游戏海报 | scenes/game.md |
| 音乐封面与艺人宣传图 | scenes/music.md |
| 概念艺术与插画 | scenes/concept-art.md |
| 摄影风格图 | scenes/photography.md |
| 抽象艺术与实验性视觉 | scenes/abstract.md |
| 活动现场视觉与展览 | scenes/event.md |
| 室内空间概念图 | scenes/interior.md |
安全使用建议
This skill appears to do what it says: convert a user’s visual brief into creative, scenario‑specific prompts using the included scene guides. Before installing, note two practical points: (1) the skill will (per its instructions) be used whenever the user mentions image-related keywords — ensure you’re comfortable with the agent autonomously invoking it in those cases; (2) it expects to reference user-uploaded images (the guidance includes explicit tags like 【图生图：商品图】 and instructions to preserve product/logo details), so avoid uploading images with confidential or third‑party personal likenesses you don’t have rights to. There are no hidden downloads, credential requests, or external endpoints in the package. If you want extra assurance, ask the publisher how the skill will be triggered in your environment (e.g., whether you can disable automatic invocation on certain prompts) and confirm any internal privacy/legal checks for using user-supplied reference images.
功能分析
Type: OpenClaw Skill
Name: limtdesign
Version: 1.0.0
The skill bundle contains metadata and extensive markdown files that provide detailed instructions for an AI agent to generate visual creative prompts. The instructions are purely textual, guiding the AI on visual styles, color palettes, composition, and handling of reference images for various scenarios (e.g., e-commerce, social media, art). There is no evidence of malicious code execution, data exfiltration, persistence mechanisms, or prompt injection attempts with harmful objectives. The handling of user-uploaded images is explicitly for image generation, not exploitation. All file access is confined to reading markdown files within the skill bundle itself, which is a standard and expected behavior for loading skill definitions.
能力评估
Purpose & Capability
The skill's name/description (visual creative prompt generation) align with the provided SKILL.md and the extensive scenes/* guidance files. It does not request unrelated binaries, credentials, or config paths. The scene files are directly relevant to producing targeted prompts for many visual scenarios.
Instruction Scope
Runtime instructions direct the agent to analyze the user's brief, identify a scenario, and read the appropriate scenes/* subfile to produce design rationale and final prompts. This is coherent for the stated purpose. It also prescribes specific reference‑image tags (e.g. 【图生图：商品图】) when using user-uploaded images — expected for an image prompt generator but something to be aware of because it will incorporate user-supplied images into generated prompts. The instructions do not tell the agent to read arbitrary system files, contact external endpoints, or access environment variables.
Install Mechanism
This is an instruction-only skill with no install spec and no code files; nothing is downloaded or written to disk. That is the lowest-risk install model and matches the skill's nature.
Credentials
The skill declares no required environment variables, no credentials, and no config paths. Its instructions reference only the bundled scene files and (optionally) user-uploaded reference images — which are proportional to its function.
Persistence & Privilege
Flags show always: false and default autonomous invocation allowed; neither is unusual for a skill of this type. The skill does not request permanent system presence or attempt to modify other skills or global agent 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 limtdesign - 安装完成后，直接呼叫该 Skill 的名称或使用
/limtdesign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
visual-creative skill 1.0.0
- Initial release of a skill for generating AI image prompts for all types of creative visual materials
- Supports posters, banners, product images, social media graphics, concept art, branding visuals, and more
- Features a structured creative framework including concept methods, composition strategies, color systems, texture, and style guides
- Includes detailed prompt output format with modular sections: subject, environment, composition, lighting, style, and text elements
- Provides a comprehensive process from analyzing user needs to transforming them into creative, high-quality AI image prompts
元数据
常见问题
limtdesign 是什么？
视觉创意生图提示词生成skill。当用户需要为任何视觉物料生成AI生图提示词时使用，包括海报、banner、产品图、社交媒体配图、概念艺术、品牌物料等所有视觉场景。无论用户的需求是模糊的（"帮我做张海报"）、半清晰的（"科技风新品发布海报"）还是已有方向的，都应触发此skill。当用户提到生图、出图、画图、提示词... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 325 次。
如何安装 limtdesign？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install limtdesign」即可一键安装，无需额外配置。
limtdesign 是免费的吗？
是的，limtdesign 完全免费（开源免费），可自由下载、安装和使用。
limtdesign 支持哪些平台？
limtdesign 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 limtdesign？
由 hitlmt（@hitlmt）开发并维护，当前版本 v1.0.0。
推荐 Skills