Launching today
Agent-Bridge
Miro-style canvas for managing AI agents and terminals
1 follower
Miro-style canvas for managing AI agents and terminals
1 follower
Self-hosted Miro-style canvas for AI coding agents and terminals across multiple machines. Each machine becomes a node with its own board. Sessions are persistent — survive browser closes, reconnect from anywhere. Status (busy/idle/error) visible at a glance via minimap and icons. Not a terminal replacement — a visual layer on top. Works with Claude, Codex, Kilo, Qwen, any CLI. Open source, MIT.
Hey 👋
I'm Vadim. I've been in software for a long time — startups, product teams, infra, the whole thing.
Lately I've been running AI coding agents (Claude Code, Codex, Kilo, Qwen) across multiple machines and hit two problems at once:
1. Isolation — give the agent full root and pray, or spend your day granting permissions. Containers help but then you're SSH-ing everywhere.
2. Visibility — sessions die between tabs, no overview of what's running where, no sense of control.
Agent-Bridge solves both. It's a canvas in the browser — each machine is a node with its own board. Sessions are persistent. Status is visible at a glance.
The setup I use daily: Incus containers on one server as isolated dev environments. Each container has docker inside, the agent gets full root but stays sandboxed. AB gives each one its own canvas board — switch between them in one click.
Built it for myself. Self-hosted, open source, MIT.
GitHub: https://github.com/agent-bridges/ab
Site: https://agent-bridge.dev
Demo: https://youtu.be/ei4Jf2JPSSk
Would love feedback — especially from anyone managing multiple AI agents across machines.