Summary

ClawTeam-OpenClaw addresses the isolation of individual AI coding agents by enabling multi-agent swarm coordination. It allows agents to self-organize, split tasks, communicate, and converge on results autonomously, with OpenClaw as the default agent. This empowers users to delegate complex goals to an agent swarm, reducing human micromanagement.

How It Works

The framework operates on an agent-spawns-agents model. A leader agent initiates worker agents via clawteam spawn, each provisioned with an isolated environment including a dedicated git worktree and tmux window. Agents communicate through auto-injected CLI commands for task updates and inter-agent messaging via inboxes. Coordination is managed through a shared filesystem state (~/.clawteam/) and visualized via terminal or web dashboards.

Quick Start & Requirements

• Prerequisites: Python 3.10+, tmux, and at least one CLI coding agent (e.g., OpenClaw, Claude Code).
• Installation: Clone the repository (git clone https://github.com/win4r/ClawTeam-OpenClaw.git), navigate into it, and install locally (pip install -e .). Avoid PyPI's upstream version.
• Setup: Requires creating a ~/bin/clawteam symlink, installing the OpenClaw skill, and configuring OpenClaw's exec approvals to "allowlist" mode. An automated installation script is available.
• Recommended: Prompt your primary agent with a complex task (e.g., "Build a web app using clawteam").

Highlighted Details

• Agent Self-Organization: Agents autonomously spawn workers, split tasks, and coordinate via CLI prompts and inboxes.
• Workspace Isolation: Each agent operates within its own git worktree, ensuring no merge conflicts and enabling robust checkpointing/merging.
• Task Management: Features a kanban-style board for task tracking (pending, in_progress, completed, blocked) with dependency resolution (--blocked-by).
• Inter-Agent Communication: Supports point-to-point messaging via inboxes and broadcast messages, with file-based or ZeroMQ transport options.
• Monitoring: Provides terminal (board show), live (board live), tiled tmux (board attach), and Web UI (board serve) dashboards for swarm visualization.
• Team Templates: TOML files define reusable team archetypes for specific domains (e.g., AI Hedge Fund).
• OpenClaw Integration: Deeply integrates OpenClaw as the default agent, transforming it into a powerful multi-agent platform.

Maintenance & Community

The project actively syncs upstream fixes and has a clear roadmap including planned features like Redis transport and shared state layers. Community links (Discord/Slack) and specific contributor details are not provided in the README.

Licensing & Compatibility

The project is released under the MIT License, permitting free use, modification, and distribution, including for commercial purposes.

Limitations & Caveats

The "Per-Agent Model Assignment" feature is currently in preview on a separate branch. Default cross-machine support is planned but not yet implemented, relying primarily on local filesystem and tmux infrastructure. Setup requires careful attention to prerequisites and configuration steps.