JARVIS: The Unified Command Center for Multi-Machine AI Coding Agents

JARVIS is a desktop application built with Tauri for orchestrating AI coding agents like Claude Code across multiple machines. It addresses the pain points of distributed AI coding management and supports features such as real-time monitoring, task distribution, Git workflow management, and pipeline orchestration, with local network deployment via SSH/Tailscale.

As AI coding assistants become more capable, developers face management challenges when deploying tools across multiple machines (local workstations, remote servers, dedicated GPU machines): they need to SSH into each machine to check status, context is easily lost when switching machines, and fragmented management leads to low efficiency, task duplication, or omissions. JARVIS's core value lies in providing a unified dashboard to centrally manage AI agents across all machines.

JARVIS covers the full lifecycle of distributed AI coding management:

1. Machine registration and health monitoring (add local/remote machines, perform health checks, collect system/GPU status);
2. Real-time agent activity stream (parse Claude Code's JSONL session files and display AI activities);
3. Task distribution and tracking (send tasks to any machine and track execution processes and results);
4. Git workflow integration (GitHub CLI supports branch management, commit tracking, and PR workflows);
5. Pipeline orchestration (define multi-step pipelines across machines);
6. Collaborative planning mode (human-machine "ping-pong" conversational collaboration to plan complex tasks);
7. Activity timeline analysis (record and visualize AI work patterns).

The tech stack balances performance, security, and user experience:

• Backend: Rust + Tauri2 (memory-safe, lightweight and secure), including modules for configuration management, machine registry, session monitoring, task distribution engine, planning mode, pipeline orchestration, GitHub integration, etc.;
• Frontend: Svelte5 + TypeScript (reactive, suitable for real-time updates), with feature tabs, command bar, and activity stream display;
• Communication mechanism: Local machines execute via shell exec; remote machines connect via SSH (supports Tailscale), with real-time return of stdout/stderr output.

Optional Integrations:

1. WhatsApp Bridge (port 3142): Receive mobile notifications and send commands via WhatsApp;
2. Voice Agent (port 3144): Voice interaction feature based on LiveKit. Configuration Deployment: Fully based on TOML files. A default configuration is automatically created on first launch. You can dynamically add/modify machines (ID, name, host address, OS, role, etc.) and code repositories without recompilation—changes take effect immediately.

Applicable Scenarios:

• Multi-device developers (unified management of AI sessions across devices);
• Team collaboration (share machine configurations and coordinate AI resources);
• GPU resource management (monitor GPU status and schedule computing tasks);
• Remote work (securely manage remote machines via Tailscale). Comparison with Similar Tools: No cloud required (data privacy), native desktop experience (Tauri advantage), deep integration with Claude Code, multi-machine orchestration capabilities (not just a simple SSH client).

JARVIS represents the evolution direction of AI-assisted development tools: from single-agent to multi-agent orchestration, from simple dialogue to complex workflow management. It addresses the practical pain points in distributed AI coding scenarios and provides a "task control center". As AI coding assistants become more capable, such orchestration tools will grow in importance. JARVIS demonstrates how to integrate AI into development workflows to maximize human-machine collaboration efficiency, and is recommended for developers using Claude Code across multiple machines.