AI Maestro Open-Source Platform Orchestrates AI Agents Across Machines

Summary

AI Maestro is an MIT-licensed platform designed to coordinate persistent teams of AI agents across multiple computers. Its dashboard can launch and monitor Claude Code, Codex, Grok Build, and terminal-based agents, while a peer mesh makes agents on different machines visible without requiring a central server. The Agent Messaging Protocol enables signed, prioritized messages and push notifications between agents. Optional persistent memory indexes conversations, source passages, entity relationships, and code-graph data so agents can retain context across sessions; secrets are kept in local vaults and redacted from output. Agents can run through local tmux sessions, Docker, AWS EC2, or AWS ECS Fargate, with headless workers for remote machines. The platform also includes gateways for Slack, Discord, email, and WhatsApp, team coordination with Kanban boards, interactive agent-generated pages, scheduled work, and approval cards. Its plugin provides 13 skills, including memory search, graph queries, agent management, messaging, secrets, and database access. AI Maestro can be installed on macOS and Linux, or through WSL2 on Windows, and requires Node.js 18+ and tmux; the dashboard runs locally at port 23000. The project states that it collects no telemetry, requires no account or hosted tier, and is free for commercial use under the MIT license.