Automation

What it measures

Four weekly shares — what fraction of merged PRs are bot-authored, what fraction of human PRs merged with no human review, what fraction of human PRs were AI-assisted, and what fraction of bot PRs closed unmerged — plus a list of the bots doing the most merge volume.

Why it matters

Automation is leverage, but it also moves work past human eyes. These shares make the trade-off visible: how much of the stream is automated, and how often human changes ship unreviewed.

How to read it

• Bot-authored share counts PRs opened by bot accounts (Dependabot, Renovate, etc.) against all merged PRs.
• No-human-review counts only human-authored PRs that merged with zero human reviews — bot auto-merges are excluded, since they are expected.
• AI-assisted share counts human PRs whose commits carry an AI co-author trailer.
• Bot PRs closed unmerged is the share of bot PRs that finished each week without merging — it includes bots closing their own superseded PRs (a newer version replacing an older one), so a high share is often normal version churn, not neglect.
• The bot list is the one place names appear — these are accounts, not people.

Anti-blame

Every share is a team-level percentage; no human is ever named. The leaderboard names only bot accounts, which represent automation, not individuals.

Common misreads

A high AI-assisted share is not good or bad on its own, and a high bot-authored share usually just reflects active dependency automation. Read the no-human-review line as the risk signal.