1688 Store Diagnosis

Specific Problem

When operating a store on the 1688 platform, data is scattered across different backends like the Business Advisor (Sycm) and Workbench. Operators often need to manually log in, copy-paste data, and generate diagnostic reports based on experience. This process is time-consuming, error-prone, and can lead to analysis biases due to data omissions or formatting issues. Traditional methods lack automation, making frequent store health checks inefficient, especially when quickly identifying issues in traffic, transactions, or inventory turnover.

How the Skill Works

This skill uses a set of pre-installed Python scripts to directly connect to 1688's internal APIs, automating the entire workflow from data collection to report generation. Core capabilities include:
- API-Direct Data Collection: Utilizes api_client.py to directly call Business Advisor's GET interfaces and Workbench's MTOP POST interfaces, bypassing browser plugins for stable and fast data retrieval.
- Intelligent Cookie Management: Employs cookie_manager.py to store and manage login credentials by store name, supporting multi-store switching, automatic expiration detection, and re-login triggering to avoid repetitive operations.
- Structured Report Generation: generate_diagnosis.py reads collected JSON data, generates a Markdown diagnostic report based on a predefined template covering dimensions like traffic, transactions, and inventory, and automatically converts it to HTML for easy sharing.

Key workflow steps are as follows:
1. Environment Setup: Execute the initialization script setup.sh, which automatically downloads dependencies, unpacks an isolated Chromium browser, and creates necessary directories for cross-platform compatibility.
2. Data Collection: Run fetch_data.py, which checks if existing Cookies are valid; if not, it launches the browser to extract new Cookies, then directly connects to APIs to fetch full data and save it as JSON.
3. Report Generation and Validation: Use generate_diagnosis.py to compile JSON data into metrics and generate a Markdown report; then run check_utf8.py to verify file integrity, and finally convert to HTML format via md_to_html.py.

Applicable Boundaries and Notes

• Environment Requirements: The skill depends on Python 3.12 and an isolated Chromium browser. On Windows, it must isolate the user's daily Chrome instance, controlling it via CDP port 9222 to prevent data leaks or conflicts.
• Mandatory Script Usage: All operations must strictly follow the script workflow; writing custom code to substitute is prohibited. Data sources are limited to API responses—fields that cannot be obtained (e.g., some real-time metrics) will be marked in the report for manual supplementation, without guessing or fabricating.
• Strict Report Template: The diagnostic report format is fixed, including chapter structure, table column names, and conclusion formats. The model can only dynamically write analyses based on data, without adding or removing chapters or decorative elements to ensure consistency.
• Complex Login Process: Browser login involves Cookie extraction and process management, requiring continuous polling within the same session to handle QR code refreshes and timeouts. It is suitable for users with technical backgrounds in controlled environments.