
Built an open-source monitor for AI coding agents on Linux — what features would you want?
Repo here: https://github.com/jomvick/Orbitos-island
MIT license, contributions welcome.

Repo here: https://github.com/jomvick/Orbitos-island
MIT license, contributions welcome.
Been building Orbitos Island, a floating desktop overlay for monitoring AI coding agents on Linux. Wanted to share some Tauri v2 specifics that took me a while to figure out. The main challenge: transparent window + click-through** The app is a fullscreen transparent window (always on top) that should only capture mouse events on the floating bar itself — not block the rest of the screen. Happy to answer questions about the Tauri setup. Repo in the comments.
Been running multiple AI coding agents simultaneously (Claude Code, Codex,
Gemini) and realized there's no good way to monitor them on Linux without
constantly switching terminals.
Built a floating overlay that shows live agent status, handles permission
prompts, and lets you jump back to the right terminal instantly.
Curious what the people here would actually want in a tool like this:
- Real-time token usage tracking per agent?
- Cost estimates?
- Alerts when an agent is stuck or waiting?
- Multi-agent view when running several at once?
- Something else entirely?
(Link in the comments)