Overview
This cross-border commerce knowledge base packages seller guidance as agent skills. Its optional runtime adds evidence imports and approval-oriented operating workflows alongside the instructional material.
Who it suits
Cross-border seller teams wanting structured operating knowledge, or technical teams prepared to configure an evidence runtime around real Amazon and Shopify data.
Who it may not suit
Operators expecting the demonstration tenant’s figures to represent their business, or anyone seeking automatic external writes without the documented approval process.
Seller use cases
- Use a seller knowledge skill to structure a decision and identify the evidence still needed.
- Import supported business, advertising, or inventory reports into the runtime to keep analysis attached to its source.
- Review a proposed operating action and its evidence before passing it through the runtime’s approval and execution stages.
What you can do
What you need
- Claude Code for the plugin; optional runtime needs Python and separately configured data connections.
Install
Claude Code
/plugin marketplace add kangise/ecommerce-ai-skills
/plugin install ecommerce-ai-skills@ecommerce-ai-skills Installation source ↗Permissions & data
Skills read local material. The optional runtime stores evidence in SQLite and may contact configured Amazon, Shopify and model-provider APIs.
Project info
- Maintainer
- kangise
- GitHub stars at check date
- 109
- Last commit
- 2026-09-18
- License
- CC0-1.0
- Last checked
- 2026-10-30
Pricing at a glance
The knowledge base is published under CC0. Running its optional Python operations layer requires infrastructure, configured data access, and model-provider usage; the public material does not include those accounts or usage allowances.
Open sourceGood to know
The knowledge plugin and optional runtime are different installations. Demo-seeded records are examples; real analysis needs imported or authorized evidence, and the runtime separates proposals, approval, and execution rather than treating a recommendation as completed work.
Alternatives to Ecommerce AI Skills
40RTY Shopify Admin Skills New
This community collection describes Shopify operating workflows, from catalog checks to order analysis. It supplies instructions and routines for an agent with Shopify access.
Shopify40RTY focuses on concrete Shopify operating routines; Ecommerce AI Skills spans cross-border knowledge and multiple data sources.
Nexscope Ecommerce Skills New
Nexscope packages seller planning and analysis instructions across several marketplaces. The collection includes listing, pricing, growth and research frameworks that work with information you provide.
AmazonShopifyeBay+3Nexscope offers task-specific seller planning skills; this project also provides an optional persistent evidence and approval runtime.
Amazon SP-API MCP New
This community connector gives assistants access to Amazon seller reporting through SP-API. It emphasizes orders, FBA inventory, fees and settlement information using Login with Amazon authorization.
AmazonAmazon SP-API MCP is a direct seller-data connection, while Ecommerce AI Skills supplies interpretation frameworks and a broader operating layer.
Frequently asked questions
Must I run the operations UI to use the knowledge?
No. The project supports reading the material, installing agent skills, and running the optional operations layer as distinct ways to use the same knowledge base.
Is the demonstration connected to my store?
The documented demo seeds an isolated tenant with demonstration data. Real store analysis requires separate connections or imports; demo results cannot establish a business outcome.
Does an approved proposal mean it has executed?
The runtime describes proposal, approval, and execution as separate states. Check the execution result and audit record before claiming that an external operating change occurred.
Can I import reports without connecting a live API?
The optional runtime documents supported CSV and XLSX evidence imports. Keep the report’s original scope and date so later conclusions can be traced to the actual input rather than assumed live data.
Checked against the official site on 2026-10-30 · Official source ↗
Sources
- Project repository · 2026-10-30
- README at reviewed commit · 2026-10-30
- Project documentation / source: LICENSE · 2026-10-30