← Back to Skills Marketplace
类目 AI 改图适配度分析器
by
liwenzhen1108-png
· GitHub ↗
· v1.0.0
· MIT-0
40
Downloads
0
Stars
1
Active Installs
1
Versions
Install in OpenClaw
/install category-ai-fitness
Description
输入三级类目表 → 自动抓取 Amazon/Walmart TOP 商品主图 → Claude 多模态分析 → 输出类目级决策表（AI改图适配度、侵权风险、经济性、竞争度）。触发词：类目分析、场景化适配、改图评估、类目筛选、搬运决策。
README (SKILL.md)
类目 AI 改图适配度分析器
输入三级类目表格，自动抓取双平台 TOP 商品数据 + 主图，用 Claude 多模态视觉分析，输出类目级综合决策报告。
用法
🚀 Streamlit Web UI（推荐）
cd ~/.openclaw/workspace/skills/category-ai-fitness
streamlit run app.py
启动后浏览器访问 http://localhost:8501，上传类目 Excel 文件，点击开始分析即可。
CLI 命令行
cd ~/.openclaw/workspace/skills/category-ai-fitness/scripts
# 基本用法：输入类目 Excel
python3 main.py --input categories.xlsx
# 指定输出路径
python3 main.py --input categories.xlsx --output ../output/report_20260529.xlsx
# 只跑 Amazon（跳过 Walmart）
python3 main.py --input categories.xlsx --amazon-only
# 只跑 Walmart
python3 main.py --input categories.xlsx --walmart-only
# 指定每类目抓取数量（默认 10）
python3 main.py --input categories.xlsx --top-n 15
输入格式
Excel 或 CSV，至少包含一列「类目」，支持以下格式混合：
- Amazon BSR URon.com/Best-Sellers-Home-Kitchen/zgbs/home-garden`
- Amazon 类目路径：
Home & Kitchen > Bedding > Sheets - Walmart URL：
https://www.walmart.com/cp/home/4044 - Walmart 类目路径：
Home > Bedroom > Sheets - 纯关键词：
yoga mat（自动双平台搜索）
脚本自动识别平台和格式，无需手动标注。
输出 Excel 字段
| 分组 | 字段 | 说明 |
|---|---|---|
| 基础 | 类目路径、平台、样本数、抓取时间 | 标识 |
| 改图核心 | 场景化适配度 (0-100) | TOP 主图场景图占比 + 商品形态 |
| AI 改图难度（低/中/高） | 透明/反光/人物/复杂细节 | |
| 推荐改图策略 | 全场景化 / 主图白底+副图场景 / 不改 | |
| 侵权风险 | 图片侵权风险（🔴🟡🟢） | 品牌物/水印/通用图占比 |
| 必须改图标记 | 通用图/品牌素材命中即 ✅ | |
| 经济性 | 价格中位数、月销中位数 | 抓取数据聚合 |
| 估算毛利率 | WFS 费率公式 | |
| 竞争 | 卖家密度 | 红海/蓝海标签 |
| 决策 | 决策档位 | ✅AI改图搬 / ✅白底直搬 / ⚠️谨慎 / ❌弃 |
| 决策理由 | 一句话说明 | |
| 参考图 | 代表主图×3 | 缩略图嵌入 Excel |
环境变量
| 变量 | 用途 | 是否必须 |
|---|---|---|
LINKFOXAGENT_API_KEY |
Amazon/Walmart 数据抓取（LinkFox API） | ✅ 必须 |
ANTHROPIC_API_KEY |
Claude 多模态视觉分析 | ✅ 必须 |
ANTHROPIC_AUTH_TOKEN |
同上（二选一） | 可选 |
ANTHROPIC_BASE_URL |
自定义 Claude API 地址 | 可选 |
ANTHROPIC_DEFAULT_HAIKU_MODEL |
自定义 Haiku 模型名称 | 可选 |
依赖
pip install -r requirements.txt
# 或手动安装
pip install anthropic openpyxl pillow requests pandas streamlit
Usage Guidance
Install only after reviewing or removing the Feishu webhook behavior and the bundled config.env credential. Treat any category list, product strategy, downloaded images, and generated recommendations as data that may leave your machine through LinkFox, Anthropic or a configured Anthropic-compatible endpoint, Walmart/image hosts, and the hardcoded Feishu webhook unless the code is changed.
Capability Tags
Capability Assessment
Purpose & Capability
The core workflow is coherent: it parses category inputs, fetches Amazon/Walmart product data, analyzes images with Claude, scores categories, and writes an Excel report. The concern is an automatic Feishu summary post containing category keywords, decisions, counts, and report name that is not part of the stated user-facing purpose.
Instruction Scope
The documentation discloses Amazon/Walmart fetching and Claude analysis, but it does not clearly disclose the hardcoded Feishu webhook, automatic result forwarding, or broad loading of local OpenClaw environment values.
Install Mechanism
Dependencies are ordinary PyPI packages for the stated app, but the package includes config.env with an apparent Anthropic API key and custom base URL. Even if not automatically loaded by the inspected code, bundling a credential-like secret is a material security issue.
Credentials
Network access to LinkFox, Walmart, image hosts, and Anthropic is expected for this skill. Reading ~/.openclaw/openclaw.json and importing arbitrary uppercase or provider env values is broader than necessary, and the Feishu webhook sends business-analysis outputs to a preset third-party endpoint without user control.
Persistence & Privilege
No background service, privilege escalation, destructive action, or system persistence was found. The skill writes reports, uploaded-file copies, image cache, and vision cache under its output directory, which is generally purpose-aligned.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install category-ai-fitness - After installation, invoke the skill by name or use
/category-ai-fitness - Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.0
初始发布：Amazon 类目抓取 + Claude 多模态分析 + 决策报告生成
Metadata
Frequently Asked Questions
What is 类目 AI 改图适配度分析器?
输入三级类目表 → 自动抓取 Amazon/Walmart TOP 商品主图 → Claude 多模态分析 → 输出类目级决策表（AI改图适配度、侵权风险、经济性、竞争度）。触发词：类目分析、场景化适配、改图评估、类目筛选、搬运决策。 It is an AI Agent Skill for Claude Code / OpenClaw, with 40 downloads so far.
How do I install 类目 AI 改图适配度分析器?
Run "/install category-ai-fitness" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is 类目 AI 改图适配度分析器 free?
Yes, 类目 AI 改图适配度分析器 is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does 类目 AI 改图适配度分析器 support?
类目 AI 改图适配度分析器 is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created 类目 AI 改图适配度分析器?
It is built and maintained by liwenzhen1108-png (@liwenzhen1108-png); the current version is v1.0.0.
More Skills