by 安全扫描
OpenClaw
可疑
medium confidenceThe skill's description matches its stated features, but the instructions claim persistent and networked behaviors (image upload/analysis, saving/sharing favorites, paid tiers) without specifying any storage, endpoints, or credentials — this mismatch is concerning and needs clarification before use.
评估建议
Before installing or using this skill, ask the publisher for concrete implementation details: where and how uploaded images are analyzed (local vs external API, and which host), where favorites and saved prompt libraries are stored and how sharing works, how effect-scoring is calculated, and which payment provider and API endpoints back the paid tiers. Do not upload sensitive images or secrets until you understand the storage and transmission policy. If the skill later adds an install script or ...详细分析 ▾
ℹ 用途与能力
Name and description align with the SKILL.md content: it aims to optimize image-generation prompts, support reverse-prompting from images, parameter tuning, A/B comparison, style mixing, and a favorites library. However, several claimed capabilities (image analysis, effect scoring, batch generation, paid tiers/API) typically require external services, storage, or runtime components that are not declared here.
⚠ 指令范围
SKILL.md instructs the agent to accept uploaded images, '自动分析' them, save and share a '收藏库', and perform A/B comparisons and scoring—but gives no instructions on where images or saved prompts are stored, what endpoints (if any) are used for image analysis or scoring, or how sharing is implemented. It also does not constrain where data may be transmitted, which is a privacy and coherence concern.
✓ 安装机制
No install spec or code files are present; this is instruction-only, so nothing will be written to disk or installed by the registry. That reduces immediate execution risk, but it also means implementation details (storage, network calls) are unspecified.
⚠ 凭证需求
The skill requests no environment variables or credentials, yet describes paid plans, an 'API' in the Enterprise plan, and features that normally need backend services (image processing, storage, sharing). The absence of declared endpoints/credentials is inconsistent with the claimed functionality and leaves unanswered where sensitive data and payments would be handled.
ℹ 持久化与权限
The skill claims persistent features (favorites library, saved mix schemes, sharing), but there is no declaration of config paths, storage buckets, or other persistence mechanisms. The skill does not request always:true or other elevated privileges, which is good, but the missing persistence design is a gap that should be resolved.
安全有层次，运行前请审查代码。
运行时依赖
无特殊依赖
版本
latestv1.1.02026/3/2
v1.1.0 introduces major new features for AI绘画提示词优化器： - 图片反推提示词：上传图片自动分析并生成相关提示词 - 提示词收藏库：保存、分类、搜索与分享优秀提示词 - 参数精细调优：分辨率、采样器、步数、CFG等参数细致可调 - 提示词对比测试：并排对比生成效果并自动评分 - 风格混合：多种AI绘画风格实时权重混合与推荐 - 整体优化Midjourney、SD、DALL-E等平台提示词适配
● 无害
安装命令 点击复制
官方npx clawhub@latest install ai-prompt-optimizer-cn-v1-1
镜像加速npx clawhub@latest install ai-prompt-optimizer-cn-v1-1 --registry https://cn.clawhub-mirror.com
技能文档
智能优化AI绘画提示词，支持图片反推、参数调优、对比测试、批量生成。
🆕 v1.1 新增功能
🖼️ 图片反推提示词
用户: 分析这张图片的提示词
用户: 从图片生成提示词
用户: 这张图是怎么生成的
反推功能：
- 上传图片自动分析
- 识别风格、构图、光影
- 生成相似提示词
- 支持Midjourney/SD格式
💾 提示词收藏库
用户: 保存这个提示词
用户: 收藏优秀提示词
用户: 查看我的收藏
收藏库功能：
- 按风格分类收藏
- 搜索已收藏提示词
- 一键使用收藏
- 分享收藏库
⚙️ 参数精细调优
用户: 调整分辨率
用户: 修改采样器参数
用户: 优化生成参数
可调参数：
- 分辨率 (512-4096)
- 采样器 (Euler/DPM/DDIM)
- 步数 (20-150)
- CFG Scale (1-30)
- 种子值
- 负面提示词
🔄 提示词对比
用户: 对比这两个提示词
用户: A/B测试
用户: 哪个效果更好
对比功能：
- 并排对比效果
- 差异高亮显示
- 效果评分对比
- 推荐最优版本
🎨 风格混合
用户: 混合动漫和油画风格
用户: 赛博朋克+水彩
用户: 创建风格组合
混合功能：
- 多风格权重调整
- 实时预览混合效果
- 保存混合方案
- 风格推荐
🎯 原有核心功能
📝 提示词优化
用户: 优化这个提示词：一个美丽的女孩
🎨 风格转换（10+种）
📦 批量生成
💰 付费方案
| 版本 | 价格 | 功能 |
|---|---|---|
| 免费版 | ¥0 | 每天5次优化 |
| 基础版 | ¥9.9/月 | 每天50次 + 收藏库 |
| Pro版 | ¥29.9/月 | 无限 + 反推 + 参数调优 |
| 企业版 | ¥99/月 | 批量 + API + 风格混合 |
让AI绘画更专业 🎨 v1.1
数据来源：ClawHub ↗ · 中文优化：龙虾技能库
OpenClaw 技能定制 / 插件定制 / 私有工作流定制
免费技能或插件可能存在安全风险，如需更匹配、更安全的方案，建议联系付费定制