Understanding the Recommendation Score
What the 0–100 recommendation score means, how the four dimensions combine, and how to read it when choosing a project.
Why a single score
Stars, forks, and issue counts each tell you one thing about a project, but never the whole story. A single recommendation score compresses several signals into a number you can sort by, so you can compare a popular-but-stale project against a smaller-but-active one on equal footing.
The four dimensions
Popularity normalizes star count so that a 1,000-star project scores well without a mega-project dominating everything. Activity buckets the time since the last update: updated within a month is best, a year ago is middling, longer is risky. Match measures how well the project's name, description, and topics align with your search keywords. Health combines the issue-to-stars ratio with a license bonus.
They are weighted 0.40 / 0.25 / 0.20 / 0.15 respectively — popularity matters most, but a dead project still can't win on popularity alone.
How to read the breakdown
A project scoring 75 overall with a high activity bar but modest popularity is often a better bet than a 90-score project resting on old popularity with a low activity bar. Always expand the per-dimension bars to see why a project ranks where it does.