What It Does — A Visual Layer for Agent Teams
Claude Code's experimental Agent Teams feature is powerful—you can spin up teams of agents that work in parallel, coordinated by a lead agent. But it all happens in the terminal. The claude_agent_teams_ui project fixes this by adding a comprehensive visual dashboard that runs locally on your desktop.
This isn't just another kanban board. It's specifically built to visualize and enhance the Agent Teams workflow. Tasks move automatically across columns as agents work. You can see all your projects and teams in one place, monitor token usage with a six-category breakdown, and even use it as a session viewer to analyze any Claude Code session history.
Setup — Zero Configuration Required
The app handles Claude Code installation and configuration for you. No API keys are required—it uses your existing Claude Code setup. Installation is straightforward:
# Clone the repository
git clone https://github.com/777genius/claude_agent_teams_ui.git
# Navigate to the project
cd claude_agent_teams_ui
# Install dependencies and run
# Follow the platform-specific instructions in the README
The project includes built-in UI for installing MCP servers, plugins, and skills, making it easier to extend your Claude Code setup than using the CLI alone.
When To Use It — Specific Workflows That Shine
Multi-Team Projects with Cross-Team Communication
The CLI doesn't support communication between different agent teams. This app implements a custom layer that lets agents from different teams message each other. If you're running separate teams for frontend, backend, and testing, they can now coordinate directly through a native mailbox system.
Code Review Workflows with Hunk-Level Control
Like Cursor's review flow, you can accept/reject individual code hunks, leave comments, and request changes. Agents automatically pick up your feedback. This is perfect for when you want to supervise agent work without micromanaging every line of code.
Solo Agent Monitoring
Don't need a full team? Run a single agent with the same visual interface. It auto-creates and tracks its own tasks on the board—giving you visibility into what plain Claude Code CLI is doing without the terminal noise.
Context and Token Optimization
The six-category context breakdown shows exactly what's consuming tokens at every step: system prompts, conversation history, file context, etc. This helps you optimize your CLAUDE.md and project setup to reduce token usage.
Key Differentiators from Similar Tools
Compared to Vibe Kanban, Aperant, or even Cursor, this tool is specifically built for Claude Code's Agent Teams:
- Cross-team communication: Agents from different teams can message each other
- Full autonomy: End-to-end agent workflows with visual tracking
- Hunk-level code review: Accept/reject individual changes like in Cursor
- Zero setup: Installs and configures Claude Code for you
- Free and open source: No subscription fees, runs locally
Practical Integration with Your Workflow
Once installed, you can:
- Launch agent teams directly from the UI instead of using
claude code agent-team create - Monitor multiple teams simultaneously on a single dashboard
- Review agent conversations per task with clean visual timelines
- Attach files and add comments that agents will incorporate
- Analyze past sessions to understand what worked and what didn't
The app acts as both a control panel and an analytics tool for your Claude Code usage.
Getting Started Today
The GitHub repository includes comprehensive documentation and a demo video. Since it's open source, you can customize it for your specific needs or contribute improvements. For developers already using Claude Code's Agent Teams, this provides the missing visual layer that makes the feature truly practical for complex, multi-agent projects.
