One dashboard for all your AI agents
Stop switching between terminals. Launch multiple Claude Code agents, build pipelines, and watch everything happen live — tool calls, file changes, results — all on one screen. Free and open-source.

Stop switching between terminals. Launch multiple Claude Code agents, build pipelines, and watch everything happen live — tool calls, file changes, results — all on one screen. Free and open-source.

From launching a single agent to orchestrating a team — one tool does it all.
Drag agents on a canvas, see who depends on whom. Connection lines update in real time as agents work, finish, or fail.
Chain agents together: analyst first, then strategist, then implementer. If one fails, the rest keep going.
Every tool call, file edit, and output appears instantly. Scroll through 50,000+ events without a hiccup.
Send follow-up instructions while agents work. Redirect or course-correct without restarting.
When one agent finishes, the next receives a summary — files created, tools used, key output.
Docker-ready with one command. Dual storage, event archival, plugin system. Extensible by design.
Turborepo monorepo with shared typed contracts.
apps/webOperator UI with spatial office layout, session management, run history, task board, and event timeline.
apps/serverOrchestration server with agent registry, run lifecycle, event processor, JSONL persistence, and adapter layer.
packages/shared15 event types, API contracts, WebSocket messages, and dependency graph utilities shared between both apps.
No subscriptions. No limits. Self-host and own your data.
Run on your own machine. Full control over your data and agents. No external services required.
Unlimited agents, sessions, and pipelines. No artificial restrictions — use every feature from day one.
AGPL-3.0 licensed. Fork it, extend it, contribute back. Built in the open for the developer community.
Clone, install, run. That is it.