Collaborate on AI coding sessions directly from your terminal, making AI development multiplayer.
Visit WebsiteFreeVisit Website
Tracked since2026
0 reviews trackedThe Bottom Line
Entry price
Free, no paid tier
Biggest pro
Facilitates seamless collaboration on AI coding without changing tools or sharing screens.
Biggest con
Currently limited to macOS for the host.
TL;DR - mpai
- Enables multiplayer collaboration on AI coding sessions (Codex/Claude Code).
- Teammates join existing AI conversations directly from their terminal with attributed prompts.
- Uses Tailscale for secure, private, and owner-controlled session sharing.
Pricing: Free forever
Best for: Individuals & startups
What is mpai?
mpai transforms individual AI coding sessions into collaborative experiences, allowing teammates to join and contribute to ongoing Codex or Claude Code conversations directly from their own terminals. It eliminates the need for screen sharing, transcript pasting, or re-explaining context, enabling seamless, attributed collaboration on AI prompts and responses. The tool is designed for developers and teams who use AI coding assistants like GitHub Copilot (Codex) or Claude Code and want to enhance their workflow with real-time, shared context.
By leveraging Tailscale for private device networking, mpai ensures that collaboration is secure and owner-controlled. The host explicitly shares specific AI sessions, and teammates join with their identity intact, contributing prompts that are attributed to them within the existing AI conversation. This approach maintains the native terminal experience while adding a crucial layer for team coordination in AI-driven development.
Pros & Cons
Pros
- Facilitates seamless collaboration on AI coding without changing tools or sharing screens.
- Maintains context by allowing teammates to enter ongoing AI conversations.
- Enhances security and privacy through Tailscale integration and explicit sharing controls.
- Preserves individual attribution for prompts within shared sessions.
- Easy setup for guests with no global install or separate account required.
Cons
- Currently limited to macOS for the host.
- Requires an existing Tailscale network.
- Alpha stage, not yet certified for large organizations.
Key Features
Multiplayer AI coding sessions for Codex and Claude CodeTerminal-native experience for hosts and guestsAttributed prompts from different teammates within a single AI sessionExplicit, owner-controlled session sharingPrivate device networking via TailscaleLive terminal transcript and session switchingViewer and participant roles with revocationAppend-only prompt audit trail
Pricing
Free
mpai is completely free to use with no hidden costs.
Reviews
$99Free with your review
Write a reviewReview mpai, get a free AI guide
Share your experience and we will send you Improve Your Thinking Patterns Using ChatGPT, free.
Explore More
mpai FAQ
How does mpai ensure the privacy and security of shared AI coding sessions?
mpai uses Tailscale to provide a private device network, ensuring that all communication is secure. Sharing is explicit and owner-controlled, meaning the host must intentionally share specific sessions. Invites are bound to the first Tailscale identity that uses them, and the system is designed as a coordination layer, not a remote shell or transcript cloud, keeping the host machine in control.
What are the technical requirements for using mpai, both for the host and a joining teammate?
For the host, mpai requires macOS, an existing Tailscale network, and Node.js 20+ (or Homebrew for installation). Teammates joining a session also need Node.js 20+ but do not require a global install; they can run a private npx command provided by the host.
Can mpai be used with AI coding assistants other than Codex and Claude Code?
Currently, mpai is built to discover and integrate with native Codex and Claude Code sessions. The documentation specifically mentions these two agents, indicating that support for others is not yet available or explicitly stated.
What is the difference between a 'viewer' and a 'participant' role in an mpai session?
While the documentation mentions 'Viewer / participant roles and revocation,' it doesn't explicitly detail the functional differences. However, based on the context, a participant can likely contribute attributed prompts to the AI session, whereas a viewer would only be able to observe the ongoing conversation and prompts without actively interacting.
How does mpai handle the attribution of prompts when multiple teammates are collaborating?
When a teammate contributes a prompt through mpai, their chosen name is attached to that turn within the exact agent session. This ensures that everyone involved knows who asked for what, maintaining clear attribution within the collaborative AI conversation.
Source: godfaddaai.github.io