using-agent-relay

Real-time inter-agent messaging uses a tmux-based relay daemon to coordinate agent communication. Start the relay daemon, wrap each agent with relay commands, and send structured messages through tmux channels. This enables sub-5ms latency coordination for multi-agent teams working on shared codebases.

Yes. Agent-relay supports file-based asynchronous inboxes alongside real-time messaging. Teams can use structured message patterns written to shared files, enabling coordination across agents that operate on different schedules or in environments where tmux isn't available.

Agent-relay provides direct messages between agents, broadcast messages to multiple agents, and multi-project bridging patterns. These structured message formats enable task handoffs, parallel work distribution, and collaboration workflows across Claude, Codex, and Gemini agent ecosystems.

CLI controls manage the relay lifecycle: start the daemon to begin coordination, use relay commands to send and receive messages through tmux or file-based channels, and stop the daemon when coordination is complete. Reading messages retrieves task updates from the team inbox.

Yes. Agent-relay wraps tmux to automate inter-agent messaging while integrating with existing terminal sessions. It's designed for teams already using tmux in shared codebase environments and adds agent coordination without replacing your current workflow.

Agent-relay satisfies sub-5ms latency requirements for tmux-based channels, making it suitable for real-time coordination scenarios. File-based asynchronous inboxes trade latency for reliability in distributed setups.