dispatching-parallel-agents

To run parallel agents, assign one agent to each independent problem domain so they can work concurrently without shared state. This multi-agent orchestration allows decoupled tasks to be executed in parallel, accelerating overall resolution time.

Use multi-agent orchestration when you have two or more unrelated failures or work items that can be executed without shared state. This approach enables parallel investigation and resolution by decoupling context for each distinct domain.

The best way to handle unrelated failures is dispatching a separate agent to each independent problem domain. By ensuring no shared state or sequential dependencies exist, agents can tackle the failures concurrently to save time.

No, you should not dispatch parallel agents if your tasks have sequential dependencies or require shared state. This orchestration approach is designed specifically for independent tasks that can be decoupled and executed concurrently without interference.

Parallel agent dispatching explicitly requires decoupled contexts without shared state. Each agent focuses on a distinct domain using a structured prompt architecture, ensuring independent investigation and resolution without interference between concurrent tasks.