agent-orchestration

Multi-agent orchestration coordinates a leader agent with teammates across in-process, tmux, or iTerm2 backends to execute parallel tasks. It uses task dependency graphs and inbox-based messaging to manage divide-and-conquer workflows like parallel code reviews or swarm-based execution.

Yes, multi-agent orchestration supports in-process, tmux, and iTerm2 backends for parallel agent execution. These backends allow agents to run concurrently, enabling scalable task management and pipeline orchestration across separate terminal sessions.

Task dependency graphs and inbox-based message passing enable structured inter-agent communication in distributed workflows. The leader agent recruits teammates, assigns tasks with defined dependencies, and uses inbox routing to coordinate execution order and data handoffs.

You need multi-agent orchestration when dividing complex tasks into parallel subtasks such as pipeline workflows, swarm-based execution, or modular backend operations. It applies when tasks require distributed coordination, inter-agent messaging, and scalable management across multiple concurrent agents.

Multi-agent orchestration implements safe shutdown protocols and robust error handling to manage agent termination across all backends. The leader coordinates structured shutdown sequences, ensuring teammates complete or safely abort pending tasks during pipeline execution failures.

No external dependencies are required to run multi-agent orchestration for parallel tasks. It leverages Claude Code's built-in TeammateTool and Task system to manage agent spawning, task dependency graphs, and inter-agent messaging natively across supported backends.