Web Article Classifier and Briefing

Problem it addresses

When engineers receive article links through wechat-access, the bottleneck is usually not reading but stable archiving: titles, sources, categories, and paths remain scattered, making topic-based review hard. This skill turns article collection into a local file workflow: fixed top-level categories, dynamic subcategories, consistent YYYY-MM-DD_title.md naming, and an _index.json for later lookup and briefing generation.

How it works

• Entry point: after a WeChat message contains a link or article body, the skill can call web_fetch to retrieve content.
• Classification: it assigns one of 10 top-level categories such as Tech Frontier, Product Management, and News; top-level expansion requires user confirmation, while second- and third-level categories are generated dynamically from themes.
• Local storage: articles are written as Markdown files, and _index.json stores title, category, date, source, and path.
• Briefings: it can generate weekly, monthly, and quarterly reports with category counts, 50–100-word summaries, trend analysis, and career, investment, and learning recommendations.

Boundaries

The skill depends on the WeChat message entry point and local file tooling, and it writes to a specified Windows directory by default. It fits a personal knowledge archive and review workflow, not multi-user permissions, cloud sync, or complex database indexing.