@minchoi I built Stoa, so biased answer. /batch makes the execution side better, but once you have a few agents going the human bottleneck becomes visibility and re-entry. I built Stoa for that coordination layer. https://t.co/WWaalKkjdy
@MervinPraison I built Stoa, so disclosure upfront. Worktree isolation plus memory across sessions are the right primitives; the gap I still felt was a clean multi-workspace view over real Claude, Codex, and OpenCode sessions. That’s what I’m building here. https://t.co/WWaalKkjdy
@wildmindai I built Stoa, so biased take. Once people discover parallel worktrees, the next bottleneck is supervising a pile of live sessions without tab chaos. That’s the layer I’m trying to make less painful. https://t.co/WWaalKkjdy
@bcherny I built Stoa, so biased add-on. Worktrees solve the isolation layer really well; the next pain I kept hitting was knowing which session was waiting, running, or safe to resume later. I built Stoa for that supervision layer around real CLI sessions. https://t.co/WWaalKkjdy
I built Stoa because juggling multiple AI CLI sessions in raw terminals gets messy fast.
Stoa is open-source : a local desktop console for Claude Code, Codex, OpenCode, and shell in one workspace.
https://t.co/cIiPhirX1E