Beads development stats for last few months

[maphew]

I thought it would be interesting to share some stats on how the Beads repo is ticking along so far in 2026.

Jan 1 – Apr 15, 2026 (105 days)

Monthly breakdown

  ┌────────────┬──────┬─────────────┬──────────────────┬────────────┐
  │   Period   │ Days │ Issues c/cl │ PRs c/m/unmerged │ Merge rate │
  ├────────────┼──────┼─────────────┼──────────────────┼────────────┤
  │ Jan        │ 31   │ 266 / 151   │ 302 / 231 / 59   │ 76%        │
  ├────────────┼──────┼─────────────┼──────────────────┼────────────┤
  │ Feb        │ 28   │ 277 / 340   │ 481 / 306 / 180  │ 64%        │
  ├────────────┼──────┼─────────────┼──────────────────┼────────────┤
  │ Mar        │ 31   │ 221 / 271   │ 490 / 245 / 237  │ 50%        │
  ├────────────┼──────┼─────────────┼──────────────────┼────────────┤
  │ Apr (1-15) │ 15   │ 109 / 84    │ 238 / 171 / 50   │ 72%        │
  ├────────────┼──────┼─────────────┼──────────────────┼────────────┤
  │ Total      │ 105  │ 873 / 846   │ 1511 / 953 / 526 │ 63%        │
  └────────────┴──────┴─────────────┴──────────────────┴────────────┘

Daily rates per month

  ┌────────┬─────────────────────┬─────────────────────┐
  │ Period │ Issues c/cl per day │ PRs c/m/unm per day │
  ├────────┼─────────────────────┼─────────────────────┤
  │ Jan    │ 8.6 / 4.9           │ 9.7 / 7.5 / 1.9     │
  ├────────┼─────────────────────┼─────────────────────┤
  │ Feb    │ 9.9 / 12.1          │ 17.2 / 10.9 / 6.4   │
  ├────────┼─────────────────────┼─────────────────────┤
  │ Mar    │ 7.1 / 8.7           │ 15.8 / 7.9 / 7.6    │
  ├────────┼─────────────────────┼─────────────────────┤
  │ Apr    │ 7.3 / 5.6           │ 15.9 / 11.4 / 3.3   │
  └────────┴─────────────────────┴─────────────────────┘

Observations

• Issues net-zero: 873 created vs 846 closed over 3.5 months. Triage is keeping pace.
• PR flood hit in Feb: created jumped +59% (302 → 481), reviewers fell behind. March shows the scar: merge rate collapsed to 50%, unmerged rejections peaked at 237 (likely bot/spam/duplicate culling).
• April recovery: merge rate back to 72%, rejections down to 3.3/day. Someone got the queue under control, or the firehose slowed.
• PRs outpace issues ~1.7×: typical of AI-agent contribution patterns (many small PRs per issue).
• Sustained review load: ~15 PRs/day inbound is the new normal since Feb. That's ~3.75 hrs/day at 15 min/PR, ongoing.
• PR volume spikes and merge rate dips track each other with ~2-week lag. Releases cluster where throughput recovers.

Trend

Issue creation is flat-to-declining (8.6 → 9.9 → 7.1 → 7.3 per day). PR creation stepped up in Feb and stayed there. The maintainers are coping, but Mar's 50% merge rate shows what happens when reviewer capacity lags for ~4 weeks: backlog converts to rejection rather than review.

I asked Claude to write the conversation that produced the above into prompt recipe:

• prompt-upstream-health-report.md