superpowers:dispatching-parallel-agents

To dispatch parallel agents, identify independent problem domains in your project and assign each to a separate agent with an isolated context for parallel investigation. This coordinates multiple unrelated investigations while preserving the original context for overall coordination.

Parallel agent orchestration is used to speed up debugging and multitasking when failures are unrelated across subsystems. It dispatches specialized agents with isolated contexts to run parallel investigations without shared state, ensuring cross-domain context leakage is prevented.

Use parallel agents when your project involves independent problems with clear domain separation that can be solved without shared state. If tasks require explicit agent goals and safe, self-contained prompts to prevent cross-domain context leakage, parallel dispatch is appropriate.

To prevent cross-domain context leakage when dispatching multiple agents, provide each specialized agent with an isolated context and safe, self-contained prompts. Ensuring clear domain separation and explicit agent goals keeps independent investigations focused without shared state interference.

Yes, you can coordinate multiple agents while keeping the original context intact. The orchestration approach preserves the original context for coordination while dispatching specialized agents with isolated contexts to run parallel investigations on independent problem domains.

Limitations of dispatching parallel agents include the requirement for clear domain separation and explicit agent goals. It applies only to scenarios with unrelated failures or independent tasks solvable without shared state, making it unsuitable for interdependent subsystems.