dispatching-parallel-agents

Parallel agents let you dispatch concurrent investigators to independent failing tests, each analyzing its own domain simultaneously. Instead of sequential debugging, multiple agents work in isolation and produce focused fixes for each problem area.

Use parallel agents when you have 3 or more independent failures across different test files or subsystems with no shared state. Each domain can be analyzed in isolation, so concurrent investigation accelerates resolution without cross-domain interference.

Each agent receives a scoped domain and strict constraints to prevent interference. Agents work only within their targeted problem area, produce per-domain findings and fixes, then output an integration-ready plan that ensures no conflicts before merging.

Yes. Parallel agents excel when failures span multiple independent test files or subsystems. Each agent focuses on one distinct problem domain with clear goals and boundaries, enabling efficient concurrent work without cross-domain impact.

Parallel agents are designed for 3 or more independent failures. Below that threshold, sequential investigation is more efficient. The skill applies when problems are truly independent with no shared state to coordinate.

No. Each agent outputs per-domain findings and suggested fixes. The skill guides review and integration of these fixes, handling conflict detection and producing a cohesive merge plan ready for implementation.