Taobao Rebate Assistant

Problem

Rebate tooling often breaks when a user has only a product link, a vague shopping phrase, or a withdrawal request, while platform rules, auth state, search parameters, and link generation remain fragmented. This skill collapses the Taobao, JD, and Pinduoduo rebate flow into one routed main path, preventing the model from inventing rebate codes or over-explaining script output.

How It Works

The skill uses rebate_assistant_router.js as the single entry point and routes by scenario:

• S01 Auth and Tutorials: handles rebate, tutorial, withdrawal tutorial, account balance, and confirm withdrawal, covering fixed flows such as start_auth, withdraw_prepare, and withdraw_confirm.
• S02 Link Rebates: recognizes Taobao, JD, and Pinduoduo product links, then runs platform recognition, product lookup, lease request, and rebate link generation.
• S03 Product Search: when no link is provided, extracts raw_text and an optional platform from inputs like brand + model or category + budget, then outputs product_search_intent and search_request.

The key constraint is that script stdout is returned as-is in Markdown, without added summaries, recommendations, or fabricated results on failure.

Boundaries

Current coverage is limited to Taobao, JD, and Pinduoduo product links. Withdrawals require completed WeChat authorization and the minimum amount. The search scenario only produces a search intent; generating a rebate link still requires the user to send the product link into S02.