superpowers-dispatching-parallel-agents

To distribute independent tasks to parallel agents, you partition tasks.md into distinct domains and assign each domain to a dedicated agent with explicit boundaries and clear acceptance criteria to accelerate execution.

To avoid shared state conflicts during parallel agent coordination, use a lightweight orchestration pattern with explicit boundaries, minimal shared state, and deterministic task handoffs for independent task execution.

You verify outputs from parallel agents by centralizing coordination and review, checking each agent's focused task results against clear acceptance criteria independently before merging the completed work.

You should partition tasks.md into independent domains for parallel investigation when you have a set of unrelated tasks that can be executed concurrently without shared state conflicts to reduce overall completion time.

This parallel dispatching approach is designed for independent tasks with minimal shared state and explicit boundaries, so it is not suitable for tasks with complex dependencies or heavy shared state conflicts.