ArticleTracker

ArticleTracker starts from a simple but persistent problem: most articles saved to a read-later list never get opened again.

Core idea

I want to split the act of saving into two stages. First, capture content with almost no friction. Then let AI evaluate whether the piece is actually worth reading now, based on the user's background, the information density, and the likely payoff.

Intended workflow

• Save an article from the browser with one click
• Extract the content and generate an assessment
• Rank the queue by score and relevance
• Record notes and summaries after the user actually reads it
• Sync the final result back into a long-term knowledge base

Why it matters

For me, the bottleneck is rarely access to information. It is deciding what deserves real attention. ArticleTracker is an attempt to turn that decision into a structured workflow.

Current focus

The project is currently centered on extraction reliability, user preference learning, and state consistency between the extension and the knowledge base.