wz:dispatching-parallel-agents

To dispatch independent tasks to parallel agents, assign each problem domain to a specialized agent. This isolates failures and prevents context bleed while enabling parallel execution without shared state.

The best way to orchestrate parallel execution for unrelated failures is to enforce one agent per problem domain. This provides structured prompts and guides parallel execution while avoiding cross-contamination.

You need parallel agent orchestration when you have multiple unrelated failures or tasks that can be solved independently. Dispatching them to specialized agents reduces bottlenecks and context bleed.

To isolate failures across independent problem domains, dispatch one specialized agent per domain. This approach avoids cross-contamination and ensures that failures in one area do not affect others.

You should not use parallel agents when tasks are dependent on each other or require shared state. This approach is specifically designed for independent problems that can be solved without cross-domain context.