Can AI coding stay native and still become multiplayer?
My cofounder Hudson and I already had our Macs connected with Tailscale, but our Codex and Claude Code sessions were still isolated. If one of us was 40 turns deep, the other got a summary, a screen share, or a finished commit—not the conversation itself.
We built mpai around a narrower bet: do not replace the terminal or the native agent. Let a trusted teammate enter an explicitly shared session, arrive with its real context, and add an attributed turn from their own Mac.
The public alpha works with Claude Code and managed-mode Codex. The host stays in control, and standalone Codex remains view-only by default.
Where should the trust boundary land for teams? Is an owner-controlled shared session meaningfully better than moving everyone into a hosted collaborative IDE, or does it still feel too close to remote access?


Replies
mpai
Launch-day update: we collapsed the host path into one paste:
brew install godfaddaai/tap/mpai && mpai start
That installs mpai, verifies the host, asks for both names, lets the host select exactly one session, and prints the teammate's private npx command. We're looking for 10 Mac-based pairs to time invite → first room. If you try it, report metadata only at https://github.com/godfaddaai/multiplayer-ai/issues/7 — no prompts, transcripts, names, tokens, or addresses.
mpai
Concrete v0.4.18 proof: a fresh isolated invite reached its first authorized shared room in 0.07 minutes (~4 seconds) across our two Macs. The local receipt exposes only elapsed time/count; old invites remain unmeasured after upgrades. Founder dogfood, not cohort progress—the first 10-team test is still 0/10: https://github.com/godfaddaai/multiplayer-ai/issues/7