ecommerce-material-studioSAFE

From the repository's own README, as read at the audited commit. Badges and raw HTML are left out.

# 1) Agent Skills CLI (works with 50+ agent hosts)
npx skills add jiawood2006/ecommerce-material-studio

# 2) Claude Code / any host that reads ~/.claude/skills
git clone https://github.com/jiawood2006/ecommerce-material-studio.git ~/.claude/skills/ecommerce-material-studio

# 3) Hermes Agent
git clone https://github.com/jiawood2006/ecommerce-material-studio.git ~/.hermes/skills/utilities/ecommerce-material-studio

pip install Pillow numpy scipy

# 1) detect category / style from a product photo
python3 scripts/category_detector.py --image product.png

# 2) recommend a template from category + price band + platform + brand
python3 scripts/style_matcher.py --category shampoo --platform taobao

# 3) scene-aware compositing — this is a Python library, not a CLI
python3 -c "from scripts.scene_aware_compositor import SceneAwareCompositor as C; C().composite('product.png','plan.json','out.png')"

# 4) text layer / 5) quality check / 6) platform resize / 7) delivery package
python3 scripts/text_engine.py --plan plan.json
python3 scripts/quality_check.py --plan plan.json --image out.png
python3 scripts/platform_adapter.py --input out.png --platform taobao
python3 scripts/delivery_packager.py --dir ./out

# 8) batch mode with resume
python3 scripts/batch_processor.py --input-dir ./products --plan plan.json

The instruction file, verbatim from the audited commit — this is the text the model reads, and the surface the audit's instruction layer examines. Quoted here so you can judge it without cloning anything.

---
name: ecommerce-material-studio
description: "电商素材工坊(中英双语)。用户需要生成电商主图、详情图、场景图、产品图合成、品牌叠加时使用。自动识别品类→匹配风格→场景感知合成→统一文字→自动质检→多平台适配→标准化交付。E-commerce product image studio: category detection, style matching, scene-aware compositing, text overlay, quality check, multi-platform adaptation, batch delivery."
version: 1.0.0
author: 涛哥
license: MIT
metadata:
  hermes:
    tags: [ecommerce, image-generation, product-photo, 电商, 主图, 素材, 场景图]
---

# 电商素材工坊 / E-commerce Material Studio

生成电商产品素材(主图/详情图/场景图)的一站式工具链。输入产品图片 → 自动完成品类识别、风格匹配、场景合成、文字叠加、质检、多平台适配、批量交付。

**One-stop pipeline for e-commerce product images: category detection → style matching → scene compositing → text overlay → quality check → platform adaptation → batch delivery.**

## 何时使用 / When to use

- 用户需要生成**电商主图/详情图**(如"帮我做一套剃须刀主图")
- 需要**产品图合成到场景**(白底图 → 场景图)
- 需要**品牌叠加**(Logo/保障条/卖点文字/徽章)
- 需要**多平台尺寸适配**(淘宝/快手/抖音/拼多多/京东等 7 平台)
- 需要**批量生成**多个产品素材

## 使用流程 / Workflow

```bash
# 0. 环境依赖(一次性)
pip install Pillow numpy scipy

# 1. 品类识别:输入产品图 → 识别品类/风格
python3 scripts/category_detector.py --image product.png

# 2. 风格匹配:品类+价位+平台+品牌 → 推荐模板
python3 scripts/style_matcher.py --category 个护电器 --sub-category 剃须刀 --price 169 --platform kuaishou

# 3. 场景感知合成(核心):库调用(SceneAwareCompositor 是 Python 库,非 CLI)
python3 -c "
from PIL import Image
from scripts.scene_aware_compositor import SceneAwareCompositor
c = SceneAwareCompositor()
scene = Image.open('scene.jpg'); product = Image.open('product.png')
result = c.composite(scene_image=scene, product_image=product, scene_type='lifestyle_bathroom', position=(0.5, 0.45))
result.save('result.png')
"

# 4. 统一文字叠加(处理 plan.json 里所有文字层)
python3 scripts/text_engine.py --plan output/plan.json --brand langke --scene-tone dark

# 5. 自动质检(读取 plan.json + 检查成品图)
python3 scripts/quality_check.py --plan output/plan.json

# 6. 多平台尺寸适配
python3 scripts/platform_adapter.py --input-dir ./output --platforms kuaishou --output-dir ./platform_output

# 7. 标准化交付打包(自动生成使用指南+清单+zip)
python3 scripts/delivery_packager.py --project-dir ./output --product-name "示例产品"

# 8. 批量处理(多产品,断点续传)
python3 scripts/batch_processor.py --input products.json --output-dir ./batch --prepare
```

## 模块清单 / Modules

| 模块 | 功能 | 依赖 |
|:---|:---|:---|
| `category_detector.py` | 品类识别(色调/材质→子品类) | Pillow |
| `style_matcher.py` | 风格匹配(品类+价位+平台+品牌→模板) | 无 |
| `brand_loader.py` | 品牌配置加载(多品牌/Logo选择) | 无 |
| `text_engine.py` | 统一文字引擎(z-index/避让/对比度) | Pillow |
| `quality_check.py` | 自动质检(分辨率/可读性/Logo/完整/重叠) | Pillow |
| `preference_memory.py` | 偏好记忆(跨项目复用风格) | 无 |
| `batch_processor.py` | 批量处理(断点续传/重试/报告) | 无 |
| `platform_adapter.py` | 7 平台尺寸适配(resize/crop/压缩) | Pillow |
| `delivery_packager.py` | 交付打包(使用指南+清单+zip) | Pillow |
| `layout_engine.py` | 布局引擎(物理尺寸→像素比例) | 无 |
| `scene_aware_compositor.py` | 场景感知合成(参照物尺度/透视/景深) | Pillow+numpy+scipy |

## 数据文件 / Data (references/)

- `category_templates.json` — 12 品类场景模板库(推荐场景/prompt/配色/文字风格)
- `product_profiles.json` — 产品档案库(示例:example_shaver)
- `brand_profiles/langke.json` — 示例品牌配置(朗科=示例品牌,非真实)
- `brand_config_template.json` — 新建品牌模板
- `user_preferences.json` — 偏好记忆库(模板)

## 注意事项 / Notes

- **字体**:`text_engine.py` 需要中文字体(macOS: `/System/Library/Fonts/PingFang.ttc`,Linux: NotoSansCJK,Windows: msyh.ttc)——按需修改 `FONT_PATHS_BOLD`/`FONT_PATHS_REGULAR` 常量
- **品牌配置**:用 `brand_config_template.json` 新建自己的品牌(含 Logo 路径/色系/保障条)
- **合成模式**:`scene_aware_compositor.py` 支持场景感知模式(自动算尺度)和兼容模式(固定 scale)
- 参考数据中的"朗科/LangKe"为**示例品牌**,可直接替换为自己的品牌配置

SAFEgrade B · trust 89/100 Nothing in the source contradicts what it says it does. Grade A is reserved for packages that have also passed the behavioural sandbox.

What the source does

Filesystem

declared (2 observation(s))

Network

none-observed

Shell

none-observed

Dependencies

pinned

Secrets in source

none-found

Findings (0)

No findings outside the package's declared scope.

Gates applied: no_behavioural_pass.