dispatching-parallel-agents

To run parallel agents without context contamination, dispatch one specialized agent per independent problem domain. This isolates each agent's context, prevents cross-contamination, and preserves your coordination focus while tackling unrelated tasks simultaneously.

The best way to orchestrate multiple agents for unrelated domains is dispatching them in parallel. This skill assigns one agent per domain, crafts focused prompts, preserves your coordinating context, and returns consolidated results for quick decision-making.

You should dispatch parallel agents when multiple failures or tasks are unrelated and can be tackled simultaneously without shared state. This pattern reduces total investigation time and surfaces independent insights quickly for consolidated review.

To start dispatching agents in parallel, provide a set of independent tasks and ask the system to dispatch dedicated agents. The skill identifies distinct problem domains, assigns one agent per domain, and collects concise summaries to inform your decisions.

Parallel agent dispatching is not suited for tasks requiring shared state. It is designed specifically for independent problem domains where tasks are unrelated, ensuring each agent's context remains isolated to prevent cross-contamination.

Parallel agent orchestration preserves your coordinating context by isolating each specialized agent's context to its assigned domain. This prevents cross-contamination across unrelated tasks while returning consolidated summaries for your review.