Juggling AI agents
just got simple.
One workspace. Every project. Claude, Codex, Bash — running in parallel, forking contexts, never losing state.
You're running 5 agents across 3 projects. Tabs everywhere. Context gone. State lost.
AgentJuggler groups everything by project. One app. Every terminal remembers. Every context forkable. Every session persistent.
Every project.
Every agent.
One screen.
The dashboard shows all your worktabs per project — active agents, idle terminals, status at a glance. Switch between projects with Ctrl+1-9. No hunting through tabs.
One golden context.
Five experiments.
Hit a sweet spot in your AI conversation? Fork it. Each fork is independent — try different approaches from the same starting point without losing the original. Split view lets you monitor them side-by-side.
Not a chat wrapper.
Full PTY power.
Every worktab runs a real pseudo-terminal powered by xterm.js. Run any command, see full output, use vim. Each terminal opens directly in your project directory.
- Full xterm.js emulation over WebSocket
- Session persistence — close and resume
- Color-coded: Claude, Codex, Bash
More than terminals.
Background Agents
Fire-and-forget AI tasks. No terminal needed. Check results when you're ready.
AI Meetings
Agent-to-agent collaboration. Multiple AIs discuss, plan, iterate — with an organizer keeping things on track.
Stack Actions
One-click commands tailored to your stack. Laravel migrations, npm builds, docker compose — instant.
Built in public. No BS.
Raw progress, real screenshots, honest takes. Follow the journey.
Ready to
tame the chaos?
Free. Local-first. No accounts. One download.