HKUDS/ClawTeam
ClawTeam
ClawTeam is a framework for coordinating multiple large language model agents to automate complex technical workflows. It operates as an agentic workflow automator and orchestrator that manages swarms of specialized agents using a leader-worker architecture to delegate and execute tasks.
The system distinguishes itself by providing isolated workspaces for parallel development, assigning each agent a dedicated git worktree and branch to prevent merge conflicts. It further enables the integration of external command-line tools by wrapping them into a standardized input and directory execution model, allowing custom CLI agents to function within the coordinated swarm.
The framework covers agent lifecycle management and task tracking via a shared kanban system to handle dependencies. It includes capabilities for inter-agent communication using file-based or peer-to-peer transport, profile-based configuration for model settings, and real-time swarm activity monitoring through dashboards and terminal views.
Features
- Multi-Agent Coordination Systems - Provides a framework for multiple specialized agents to collaborate and delegate sub-processes via a leader-worker architecture.
- Agent Swarms - Coordinates multiple LLM agents using a leader-worker architecture to automate complex technical workflows.
- Agent Lifecycle Management - Provides operations for creating and managing the runtime lifecycle of AI agent instances within a swarm.
- Worktree Isolation - Uses Git worktrees to provide isolated file states for parallel agent tasks, preventing merge conflicts.
- Parallel Worktree Development Sessions - Executes concurrent development sessions across multiple isolated Git worktrees to prevent conflicts.
- Programmatic Agent Spawning - Dynamically spawns new worker agents with dedicated identities and isolated worktrees to execute specific tasks.
- Agentic Workflow Automation - Operates as an automation engine that coordinates AI agents to execute complex technical workflows.
- Multi-Agent Orchestration - Coordinates the decomposition and delegation of tasks across a swarm of specialized AI agents.
- Coordinator-Worker Orchestration - Implements a coordinator-worker orchestration pattern to delegate complex workflows to specialized worker agents.
- Wrappers - Provides a standardized input and directory execution model to wrap external command-line tools into the agent swarm.
- Agent Task Dependency Resolvers - Automatically unblocks and resolves task dependencies for AI agents using a shared state tracking system.
- Kanban-Based Tracking - Tracks agent progress and automatically triggers dependent tasks through a shared kanban system.
- Git Workspace Isolations - Prevents merge conflicts by assigning each agent a dedicated git worktree and branch for parallel development.
- Agent Lifecycle Management - Manages the startup, shutdown, and idle states of autonomous agents to maintain swarm efficiency.
- Agent Task Managers - Manages software engineering tasks and agent lifecycles using a shared kanban system and execution plans.
- Agent-to-Agent Communication - Implements file-based and peer-to-peer transport mechanisms for communication between distributed agents.
- Team Bootstrapping - Bootstraps complete teams of agents with predefined roles and prompts using configuration templates.
- Agent Configuration Profiles - Saves provider and model settings into named configuration profiles for initializing agent instances.
- CLI Wrappers - Standardizes external command-line tools into a consistent input and directory execution model for agent consumption.
- Task Tracking Systems - Provides a shared kanban system for monitoring agent progress and managing technical task assignments.
- Agent Model Profiles - Provides named profiles for AI model parameters and provider settings to decouple configuration from environment variables.
- Agent and Integration Interfaces - Bridges command-line functionality with autonomous agents by wrapping external CLI tools for task execution.
- Plan Approval Workflows - Implements mechanisms for human-in-the-loop review, modification, and approval of proposed agent implementation plans.
- AI Agent Activity Monitors - Tracks agent progress and tool usage in real-time through terminal kanban boards and web dashboards.
- Autonomous Research Agents - Swarm intelligence for distributing research tasks across parallel agents.
