What you missed.

Newest first. Every finding links to its own page and to the evidence behind it.

Measured against each repository’s own median gap between releases, not against a fixed period — a project that ships weekly and a project that ships twice a year are both on time when they keep to their own rhythm. Under four releases on record there is no rhythm to be late against, and those are counted as forming rather than guessed at.

Nothing on the watchlist is more than 2.5× past its own release rhythm. 0 repositories have enough releases on record to be measured this way; 46 do not yet.

The list is OSSInsight’s, taken unaltered and credited to them — they read ten billion GitHub events and there is no point competing with that. What is added is the column a momentum ranking cannot carry: how many contributors account for half the commits. Stars gained are over three months, not a total, and not a week — a weekly window is dominated by five-day-old repositories where a bus factor of one is a fact about the calendar. Ordered by concentration rather than momentum, because the interesting case is where the two disagree.

Read from the version tag against the previous one. Calendar versions are excluded — a project on 2026.08 bumps its leading number every January, and reporting that as a breaking change would fill this page with nothing every new year. A major bump is a signal to read the changelog, not a claim that anything of yours broke.