All commands

Harness Component — Command

Claude Flow Swarm

Coordinate multi-agent swarms for complex tasks

Runtimeclaude-code
Intentbuild

Definition

🐝 Claude-Flow Swarm Coordination

Advanced multi-agent coordination system with timeout-free execution, distributed memory sharing, and intelligent load balancing.

Basic Usage

./claude-flow swarm "your complex task" --strategy <type> [options]

🎯 Swarm Strategies

  • auto - Automatic strategy selection based on task analysis
  • development - Code implementation with review and testing
  • research - Information gathering and synthesis
  • analysis - Data processing and pattern identification
  • testing - Comprehensive quality assurance
  • optimization - Performance tuning and refactoring
  • maintenance - System updates and bug fixes

🤖 Agent Types

  • coordinator - Plans and delegates tasks to other agents
  • developer - Writes code and implements solutions
  • researcher - Gathers and analyzes information
  • analyzer - Identifies patterns and generates insights
  • tester - Creates and runs tests for quality assurance
  • reviewer - Performs code and design reviews
  • documenter - Creates documentation and guides
  • monitor - Tracks performance and system health
  • specialist - Domain-specific expert agents

🔄 Coordination Modes

  • centralized - Single coordinator manages all agents (default)
  • distributed - Multiple coordinators share management
  • hierarchical - Tree structure with nested coordination
  • mesh - Peer-to-peer agent collaboration
  • hybrid - Mixed coordination strategies

⚙️ Common Options

  • --strategy <type> - Execution strategy
  • --mode <type> - Coordination mode
  • --max-agents <n> - Maximum concurrent agents (default: 5)
  • --timeout <minutes> - Timeout in minutes (default: 60)
  • --background - Run in background for tasks > 30 minutes
  • --monitor - Enable real-time monitoring
  • --ui - Launch terminal UI interface
  • --parallel - Enable parallel execution
  • --distributed - Enable distributed coordination
  • --review - Enable peer review pro
View full source (5,172 chars) on GitHub

More from ruvnet/agentic-flow