spawn

Launch parallel agents in isolated git worktrees to prevent interference. Each subagent operates in its own filesystem context, allowing independent execution of competing approaches without causing merge conflicts during concurrent task exploration.

Use parallel agent spawning to explore multiple refactoring strategies simultaneously. This dispatches N subagents to work independently on competing implementation approaches, reducing trial-and-error time by comparing results before choosing the best solution.

Session coordination loads session config, writes sequential dispatch assignments, tracks results, and updates session state. This provides a controlled workflow for managing concurrent task exploration and monitoring progress through the session board.

Yes, git worktrees keep each agent in its own isolated filesystem context to prevent interference. This isolation is useful when you want several debugging approaches or implementation strategies explored in parallel before committing to a winner.

Parallel agent orchestration is not suited for tasks requiring sequential execution or shared state modifications across agents. It is designed specifically for competing approaches in isolated contexts, so dependent or highly coupled task workflows will not benefit from concurrent dispatch.