1688 E-commerce Data Daily Report

Problem: Pain Points in Manual Operations Data Reporting

1688 sellers need to regularly review key metrics such as product performance, traffic sources, transaction data, and customer structure. The traditional method involves manually logging into the seller backend, checking Sycm (business advisor) and workbench data item by item, and organizing it into Excel or documents. This approach is not only time-consuming and labor-intensive but also prone to omissions or errors due to human oversight. For teams operating multiple stores, repetitive tasks multiply, diverting focus from data analysis and strategic optimization.

Solution: Script-Driven Automated Collection and Report Generation

This skill automates the entire workflow using a set of pre-packaged Python scripts:

• Secure Login and Cookie Management: Uses an isolated Chromium browser instance (separate from the user's daily Chrome) to complete QR code login and automatically extract and save cookies. The cookie_manager.py script manages cookies by store name and expiration, supporting seamless switching between multiple stores and avoiding frequent logins.

• API-Driven Data Collection: Core data is fetched directly via the api_client.py script calling 1688's internal APIs, avoiding reliance on browser plugins or page parsing. This includes retrieving Sycm data (e.g., traffic, transactions, keywords) from sycm.1688.com and workbench data (e.g., New Lighthouse fulfillment) from h5api.m.1688.com. Raw data is saved as structured JSON files.

• Standardized Daily Report Generation: Running the generate_report_v3.py script reads the collected JSON data and automatically generates a Markdown daily report with 8 data sections covering product basics, traffic data, trade & after-sales, new vs. returning customers, New Lighthouse fulfillment, top traffic channels, top keywords, and top-selling products. The filename includes the store name for easy differentiation.

• Intelligent Delivery and Push: Based on the usage scenario, the delivery method is automatically selected. In Feishu group chats, push_feishu_post.py is called to push formatted tables directly. In local or native chat environments, an HTML report is generated and uploaded to OSS, returning only a viewable link to avoid chat flooding.

Considerations and Applicable Scenarios

• Mandatory Browser Isolation: On Windows, the skill must use a pre-packaged standalone Chromium, isolated via a specific CDP port and user data directory. Connecting to or reusing the user's daily browser is strictly prohibited.
• Data Sections Read-Only: In the generated report, the first 8 data sections are output by scripts; the model is absolutely forbidden from modifying, deleting, or rewriting them. Only the ⚠️Data Diagnosis and ✅Tomorrow's Work Checklist placeholders are allowed to be filled with analysis content.
• Channel Dictates Delivery: The corresponding push logic must be executed based on the current conversation channel (e.g., Feishu vs. others). Incorrect selection will lead to table rendering failures or broken links.
• Operations Limited to Scripts: All steps, from data collection to report generation, must be strictly performed through scripts in the $SKILL_DIR/scripts/ directory. Writing temporary code in the chat to bypass scripts is prohibited.