Swarm Orchestration

Parallel task orchestration distributes work across multiple agents simultaneously. Initialize a swarm topology using `agentic-flow hooks swarm-init`, then use `task-orchestrate` with `--mode parallel` to execute tasks concurrently across agents, reducing execution time and enabling collaborative workflows.

Swarm orchestration supports three topologies: mesh (peer-to-peer agents communicating directly), hierarchical (queen-worker model with central coordination), and adaptive (topology adjusts based on workload). Choose based on your coordination needs and fault-tolerance requirements.

Load balancing automatically distributes tasks to prevent agent overload, while fault tolerance detects and handles agent failures to maintain continuous operation. The orchestration layer manages workload distribution and failure recovery transparently.

Yes, memory coordination enables agents to share state and context across the swarm. This allows cohesive decision-making where agents access shared information, ensuring consistency and enabling collaborative problem-solving in distributed AI systems.

No prerequisite knowledge is required. The Skill provides command-line hooks for initialization and task orchestration, abstracting complexity. Start with `swarm-init` to set up topology and agent spawning to begin coordinating parallel tasks immediately.

Building a REST API with tests in parallel exemplifies swarm orchestration: assign the coder agent to write code, tester to validate functionality, and reviewer to ensure quality—all concurrently. This pattern scales to other collaborative workflows requiring specialized agent roles.