Repository Quality Indicator for Github Software Projects

Read the Paper

Check a repository's quality score

Pulls GitHub activity, issue-response health, and popularity for a repo, scores each against fixed quality anchors and returns a single 0–5 rating with a full breakdown.

Example: https://github.com/pandas-dev/pandas

Last 5 repos rated

Repo Recency Issues Popularity Final
borgbackup/borg ★4.96 ★5.0 ★5.0
4.99
→
bioinfo-biols/CIRI-full ★0.0 ★1.84 ★1.9
1.25
→
COMBINE-lab/alevin-fry ★4.38 ★4.36 ★3.4
4.05
→
COMBINE-lab/salmon ★4.68 ★4.43 ★4.07
4.39
→
helion-fpga/helion ★4.38 ★1.0 ★1.7
2.36
→
How we calculate each dimension

Recency

We pull the timestamp of the most recent commit on the default branch from GitHub's GraphQL API and count the days since then. That count is mapped onto a 0–5 scale against fixed anchors: a commit today scores 5.0, easing down through 14, 30, 60, 90, and 180 days to 1.3 at a year and 0.0 at two years of inactivity. Repos with no commit data available default to the two-year (worst-case) anchor.

Issues

We pull every issue (open and closed) via GitHub's REST API and calculate how long each closed issue stayed open, using the median across the repo when there's enough data (falling back to the average). That day count is scored the same way as recency. Same-day closes score 5.0, easing down through 7, 14, 30, 60, and 120 days to 1.0 at a year. Repos with no closed-issue history default to the worst-case anchor.

Popularity

We add up stars, forks, and watchers, then take log₁₀ of the total so each additional order of magnitude (10 → 100 → 1,000 → 10,000+) is worth one point.

Final rating

The three 0–5 scores are averaged in equal thirds for the final rating shown everywhere on this site.

Analyzing repository…

Pulling GitHub metrics and scoring them against the anchor rating system. This usually takes a few seconds.