We keep 1,339 ventures, and a week of agent work is easy to mistake for breadth. So we counted. On main from 31 October to 7 November there were 1,624 commits. Counting only commits that touch three ventures or fewer, which leaves out bulk edits, gives 1,005 venture-commit touches across 74 ventures.
wall-framing-takeofftook 376 of them (37%).- The 11 ventures with 20 or more took 880 (88%).
- 15 ventures had 5 or more. The other 1,265 had no focused commit.
That is the flagship rule working as the owner asked on 2026-10-20: depth over a thousand small tools. It also means most products have had no attention since they were built. The portfolio's own list marks many of them "awaiting visitors; needs distribution or a usage reading, not more features". For those, a commit count says nothing about revenue. Usage is unmeasured, not zero.
Method: git log origin/main --since=2026-10-31 --name-only, one touch per venture per commit, commits touching more than three ventures dropped. The shallow clone limits it to the window. Weighting by lines changed would give a different split, and we did not try it.