Run multiple Claude agents from your local machine. Create tasks, assign models, watch them code in real-time — all from a single board.
You describe the task. Your agents handle the rest.
Describe what you need — a feature, bugfix, refactor, or test. Pick a model and priority.
Drag to "In Progress" and Claude creates a branch, reads the codebase, writes code, and commits — all visible in the live terminal.
Task moves to Testing. Review the diff, approve or request changes. Claude picks up your feedback and iterates.
Everything you need to run a squad of Claude agents from your own machine.
Watch each agent's tool calls, file edits, and commands in real-time. See exactly what your squad is doing.
Each agent creates its own branch. Optional auto-PR on completion. Full commit history per task.
Slack, Discord, Teams, or custom endpoints. Per-project config with event filtering and one-click test.
Run up to 5 agents simultaneously. Auto-queue chains tasks — when one agent finishes, the next starts.
Token tracking, cost analysis, model breakdown, 30-day trends, and rate limit monitoring.
Snippets inject project rules into every prompt. Templates with variable substitution. File attachments.
Click to explore different views.
Free & open source. Requires Claude Code CLI.
Or run from source: git clone && npm run setup && npm run dev
One board. Multiple agents. All running on your machine.
Star on GitHub