Commit

I'm curious about how others are handling this... How are you keeping CI and merges moving when several coding agents are opening PRs in parallel? I upgraded to GitHub Pro and have hardware at home, but much of that hardware is already occupied by the agent sessions themselves. I already use Octopool to cache GitHub reads across agents. CI capacity and keeping PRs moving through review and merge are still bottlenecks, especially when the home hardware is also running the coding sessions. With Brigade, each PR fans out across Python versions, test shards, and platforms. Rebases and follow-up pushes create more runs, and someone still has to handle review threads and merge gates. I’m weighing a dedicated runner pool, more selective checks, and a merge queue. I need verification and merging to keep going after the original session ends. Would a few potatoes on a shelf make useful self-hosted CI runners? GitHub Actions is becoming a bottleneck between opening PRs and merging them. I have hardware at home (to some extent) it's just spread between homelab and dev boxes. I’m wondering how much a few older machines dedicated to CI would help. If you’re doing this at home, what hardware has actually been worth using, and which jobs do you keep on GitHub-hosted runners?