07-SKILL-DISPATCHING-PARALLEL-AGENTS

To dispatch parallel agents, configure the dispatcher to analyze your inbox, identify 3 or more independent tasks without cross-dependencies, and launch separate agent processes concurrently to speed up execution.

Parallel task dispatch is used to process multiple independent files or tasks concurrently instead of sequentially, enforcing task scoping and preventing interference by running each task in isolation with no shared state.

You should use parallel agents when you have 3 or more files or tasks from different sources that can be processed in isolation without cross-dependencies, ensuring faster results through concurrent execution.

Isolated execution works by enforcing task scoping for each parallel agent, ensuring they run in their own scope with no shared state, which prevents interference and maintains independent processing environments.

Parallel agent outputs are consolidated by collecting the per-agent results after isolated execution and merging them into a single, coherent results state for final review.

A key limitation is that parallel agents require tasks to have no cross-dependencies; if your tasks rely on shared state or sequential processing, this isolated dispatch approach is not suitable.