Agent-Bridge
Miro-style canvas for managing AI agents and terminals
Agent-Bridge provides a visual control plane for managing persistent coding agents and terminals across multiple machines, akin to a self-hosted digital canvas. Key features include:
• Per-node canvas boards for separate workspaces
• Real-time session status monitoring at a glance
• Flexible layout management with rulers and grids
• Integrated node connection manager via UI
• Persistence of sessions across browser closes
This platform transforms each connected machine into its own board, allowing users to organize live terminals, coding agents, file managers, and notes within an interactive canvas. Unlike traditional terminal emulators, it serves as a visual layer that enhances oversight and control, making it ideal for juggling multiple projects and remote development environments. Users can quickly identify busy, idle, or error states for every window, icon, and minimap dot, and navigate seamlessly between different canvas points.
Agent-Bridge is built for managing agent swarms, integrating with any command-line interface tool such as Claude, Codex, Kilo, Qwen, or even `npm run dev`. It supports use cases like persistent agent work where sessions remain active for hours, remote machine integration via a simple SSH deploy wizard, and parallel multi-project development. The open-source, MIT-licensed nature ensures flexibility and community-driven enhancements, offering a robust solution for developers and teams managing distributed coding tasks.