Atlas · Details
Vibe Maintainer
Author’s note
A lot of people have reached out to tell me that this was an impactful post for them. It's a maximalist view on accepting AI-assisted contributions. Being a vibe maintainer is not everyone's cup of tea, but it is an important workflow that developers should understand. The thesis is that everyone will work this way within a year or two.
AI Notes
By the time Steve hit publish he was shipping toward 50 contributor PRs a day across Beads (20k stars) and Gas Town (13k stars) — 1,000+ unique contributors, 4,000+ PRs, 15,000+ commits, median time-to-resolution under fifteen hours. Most OSS maintainers handle that volume by banning AI submissions; Steve does the opposite, encouraging everyone to use AI and assuming 99% of incoming PRs are agent-assisted. The argument is that in 2026 every user with a coding agent is a credible fork — Roo forked Cline, Kilo forked Roo, and grandma's gardening community can route around you in a weekend if you say No too often. So his workflow optimizes for community throughput instead of contributor discipline. The PR Sheriff loop — now run by the Mayor agent, which turned out to be "crazy" better at it than the crew — categorizes every PR as easy-win, fix-merge, or needs-review, and the resolution ladder runs merge → merge-fix → fix-merge → cherry-pick → split-merge → redesign → retire → reject, with "request changes" dead last because it starves contributors against project velocity.
It's the operational counterpart to Gas Town: From Clown Show to v1.0 (the project-side retrospective written the same week) and the maintainer's-eye view of the Software Survival 3.0 thesis.
Related listings
-
2026
Gas Town: From Clown Show to v1.0
Three days later. The retrospective on the projects whose PR firehose this essay describes — Beads and Gas Town cutting v1.0 together.
-
2026
Welcome to Gas City
Three weeks later. Where the energy goes after Gas Town is feature-complete — large features get redirected to Gas City, which shrinks the maintainer's surface area.
-
2025
Software Survival 3.0
The underlying argument: software that doesn't keep up with what its users want gets routed around. Vibe Maintainer is the maintainer's-eye view of the same thesis.