How I built an Engineering Health Score (0–100) to filter out hype and fake GitHub repos

Hey makers! 👋 As a developer, I used to check GitHub Trending every morning—until I noticed how noisy it had become. Between marketing-driven star spikes, empty boilerplate repos, and spam clones, raw star counts alone don't tell you whether a project is actually safe or actively maintained for production use. Over the past few weeks, I built GitRank (launching here tomorrow!) to solve this exact problem. Instead of just ranking by stars, we evaluate repositories using a multi-dimensional Engineering Health Score (0–100) based on: - Commit cadence & recent maintenance signals - License clarity & repository structure - Real 7-day star velocity vs. historical baseline - Automated filtering of low-quality/spam repos We also package the top movers into a daily editorial briefing and weekly digest so developers can catch real breakthroughs in 3 minutes. I'd love to hear from fellow builders here: When you evaluate an open-source dependency for a production project, what's the #1 red flag or signal you look for first? (If you're curious to test it on your own repo, GitRank goes live on the Product Hunt front page tomorrow—any early feedback is hugely appreciated!)