AI Cross Border Ecommerce Agent Toolkit skill
Seven workflows for a Shopee and Mercado Livre seller: category research, pricing, listing, images, video, ads and review analysis
- 977 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- The skill runs Python scripts over your category, ad and review exports and reads those files
- Exports can contain store and buyer data, so avoid handing such files over without need
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add skills/ai-cross-border-ecommerce-agent-toolkitDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- The skill runs Python scripts over your category, ad and review exports and reads those files
- Exports can contain store and buyer data, so avoid handing such files over without need
This is third-party code. Review the repository files before installing.
What it does
The skill turns a cross-border selling playbook into seven processes that an agent runs step by step. The root SKILL.md acts as a router: from the user's request it opens the right submodule, and each submodule has its own rules, prompts, input templates and a worked example on a real product. The first two modules do the arithmetic. Category research parses a marketplace export with scripts: market capacity, seller concentration, price bands and spec combinations. Product strategy calculates net margin, break-even price and coupon headroom for every SKU, with fee presets for five marketplaces. The agent does not guess figures: if a field is missing from the export it says so and asks for data. Listing, images and video follow. For the listing, scripts measure title length and check parameters against the spec table; for the main image they check pixels; for video they check duration and subtitles. The loop closes with an ad funnel diagnosis in four types and a review analysis in eight causes. Their findings flow back into the listing and images, and the fixes are re-tested. Buyer-facing text is written in Brazilian Portuguese.
Who it is for. For marketplace sellers and managers who run stores on Shopee Brazil and Mercado Livre and want an agent to deliver steady results by a fixed playbook.
Good fit when
- Before launching a product you want to assess the category from an export: capacity, seller concentration and price bands
- You need to price a product and design its SKUs with the marketplace fee, coupons and a target margin in mind
- You need a Portuguese title and description with a length check and parameters verified against the spec table
- Ads burn budget and you want a funnel diagnosis with a profitability line and ranked actions
- Negative reviews have piled up and you want them sorted by cause and turned into listing fixes
Not a fit when
- You need actions inside the seller dashboard: the kit changes no prices or budgets, it hands you an action list to approve
- You want fake orders, solicited reviews or false reports against competitors: the kit refuses such requests
- You need data collected from the marketplace: you bring the category, ad and review exports yourself
- You need a finished image or video: the kit produces a plan and prompts, and other tools draw and edit
Example request
Analyze the electric scooter category from this export: capacity, seller concentration and price bands, then propose a SKU price including the Shopee feeLimitations
The kit is tuned for Shopee Brazil and Mercado Livre. For Amazon, TikTok Shop and Shopify only the fields are adapted. Marketplace rules are as of writing and must be rechecked in the seller dashboard before publishing. The pricing model leaves out Brazilian taxes, so you add the rate yourself.
You provide the category, ad and review exports; the kit collects no data. It does not create images or video, only a plan and prompts, and it does not act in the dashboard or change prices and budgets.
The docs folder holds the original 38-page internal playbook unedited. It describes practices the marketplaces forbid, and the modules do not implement them. Project status is beta: the repository description says 0.3 and SKILL.md says 0.4.0.
How to disable. Delete the kit's folder from the agent's skills directory and open a new session. The skill creates no files outside this folder; results sit wherever you saved your exports and texts.
FAQ
Which marketplaces does the kit support?
Shopee and Mercado Livre are fully supported: title limits, main image rules and Portuguese output. For Amazon, TikTok Shop and Shopify only the fields are adapted, without detailed rules.
Do I need an API key?
No. No key or marketplace connection is needed: the kit's text files and the exports you bring are enough. The scripts need Python, and ad_metrics uses only the standard library.
What do the scripts calculate and what does the agent do?
Scripts calculate market concentration, price bands, net margin, break-even price, title length and review statistics. The agent interprets the result and writes the text. The kit's rule is that anything computable is never left to the model's guess.
How does the closed loop work?
The ad diagnosis and review analysis point to the part of the listing or images that fails. The agent fixes only that part, then the result is re-tested. Without improvement the kit goes back to diagnosis rather than doubling down on the same action.
What about practices the marketplaces forbid?
Fake orders, solicited reviews, false reports and inflated parameters are not implemented. A Compliance Guard document gives allowed replacements, and a separate register lists the playbook items left out.
How do I install the kit in Claude Code?
Clone the repository with git clone and put the folder into ~/.claude/skills. The root SKILL.md routes requests to the submodules, so you do not install them one by one.
Discussion
No questions or reviews yet.
Related
SkillsE-commerce Visual Copywriting
ecommerce-visual-copywriting
A skill that turns product data into a plan for main images, the product page and A+ content, with on-image copy, generation prompts and a claims check
A set of agent skills for affiliate marketing: program search, trend research, articles and posts, landing pages and analytics
Nine skills for taking a product overseas: Meta, Google and TikTok ads, cold email, UGC, social media and a 90-day strategy.
Marketing Skills for AI agents
Marketing Skills for AI Agents
About fifty skills for CRO, copywriting, SEO, analytics, ads and launches, built for coding agents
The clipcat skill and CLI help a TikTok Shop seller find videos that sell and produce similar ones with their own product
Marketing & SEO Skills
Marketing & SEO Skills for AI Agents
A library of hundreds of marketing and SEO skills for agents: content, page types, channels, ads and strategies
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Repository stars | Main difference |
|---|---|---|---|
| AI Cross Border Ecommerce Agent Toolkit (this card) | no | 977 | |
| clawEC cross-border trading agents | no | 72 | agents for Amazon, TikTok, Shopee and Ozon focused on product selection |
| LinkFox Skills | no | 119 | 153 skills over the paid LinkFox API, with data taken from the service |
| Ozon Product Card Studio: listing research | no | 0 | prepares Ozon listings from competitor research |
README in short
The Chinese description explains that the kit is a method, prompts, templates and examples, neither an app nor a service. It goes on with a table of the seven modules and their versions, a chain diagram from category research to review analysis, the folder layout and the supported marketplaces. It closes with the project principles, notes on marketplace compliance and a roadmap to version 1.0.
README badge
Original SKILL.md
--- name: sop-ecom-toolkit description: >- 跨境电商运营 AI 工具箱总入口（Shopee 巴西站 / Mercado Livre 专精，PT-BR）。 覆盖 7 条工作流：市场调研、产品策略与定价、Listing 文案、商品图（9 图框架）、 短视频物料、广告诊断、评论分析，以及"诊断→回改→复测"闭环。 当用户要"看类目 / 定价 / 写标题 / 写详情页 / 做主图 / 做详情图 / 9 张图 / 做视频 / 分镜 / 配音字幕 / 广告诊断 / ROAS / ACOS / 差评分析 / 卖点提炼 / 优化转化"时使用。中文「做图 / 做视频 / 看类目 / 算价格 / 分析评论」同样触发。 NOT for 账号操作、刷单测评等平台禁止行为、纯 Excel 透视。 allowed-tools: - Bash(python *) - Read - Write metadata: version: 0.4.0 category: e-commerce/operations license: MIT --- # AI Cross Border Ecommerce Agent Toolkit（总入口） 本文件夹是**伞形 Skill**：先按下面的路由表判断用户要做哪件事， **再打开对应子模块的 SKILL.md 严格照做**——子模块里有各自的规则卡、 算术脚本与底线约束，本文件只负责路由，不替代子模块的细节。 ## 全局三条底线（任何子模块都继承） 1. **不编数据** —— 输入缺字段报 `[INPUT_MISSING]` 并列出可用列，不许猜； 指标一律跑 `scripts/` 下的脚本，Agent 不心算 2. **合规优先** —— 刷单/诱导评价/虚假宣传/恶意举报等平台禁止行为不做 （见 `docs/03_Compliance_Guard.md`、`docs/05_SOP_Gap_Register.md`） 3. **买家可见文本 = 目标市场母语（PT-BR）**，内部备注用中文； 所有数字必须溯源到 `spec_table` ## 路由表 | 用户要做的事 | 打开哪个子模块 | |---|---| | 看类目 / 市场容量 / 集中度 / 价格带 / 配置机会 | [Agents/01_Market_Research_Agent/SKILL.md](Agents/01_Market_Research_Agent/SKILL.md) | | 定价 / SKU 结构 / 优惠券额度 / 目标毛利反算 | [Agents/02_Product_Strategy_Agent/SKILL.md](Agents/02_Product_Strategy_Agent/SKILL.md) | | 标题 / 详情页 / 卖点提炼 / 关键词分类 / 合规扫描 | [Agents/03_Listing_Creative_Agent/SKILL.md](Agents/03_Listing_Creative_Agent/SKILL.md) | | 主图 / 白底图 / 场景图 / 9 张图 / 图合规检查 | [Agents/04_Product_Image_Agent/SKILL.md](Agents/04_Product_Image_Agent/SKILL.md) | | 视频 / 分镜 / AI 视频 Prompt / 配音 / 字幕 | [Agents/05_Product_Video_Agent/SKILL.md](Agents/05_Product_Video_Agent/SKILL.md) | | 广告诊断 / ROAS / 盈亏线 / 预算动作 | [Agents/06_Advertising_Optimization_Agent/SKILL.md](Agents/06_Advertising_Optimization_Agent/SKILL.md) | | 差评归因 / 评论统计 / 卖点排行 / 修补文案 | [Agents/07_Review_Analysis_Agent/SKILL.md](Agents/07_Review_Analysis_Agent/SKILL.md) | **跨模块闭环**（用户在做完整运营时提醒他这条链路）：