dispatching-parallel-agents

Parallel debugging dispatches separate AI agents to investigate independent test failures simultaneously rather than sequentially. Each agent focuses on a specific failing test file, reducing overall debugging time and enabling concurrent problem-solving across unrelated issues.

Use parallel agents when you have 3 or more independent failures across different files or subsystems with no shared state. This approach eliminates sequential investigation overhead and is ideal after major refactors that break multiple unrelated components.

Yes. Parallel agents work on distinct, independent issues simultaneously, making them effective for multiple subsystem breakdowns or unrelated bugs. Each agent receives focused scoping and clear goals for its domain, enabling concurrent resolution without shared state conflicts.

After concurrent investigation and fixes, a post-dispatch integration review consolidates the results across all domains. This ensures that individually resolved issues integrate correctly and maintain system coherence.

Yes. This skill applies only to independent failures with no shared state. If failures depend on common root causes or share state, sequential investigation or unified diagnosis is more appropriate than parallel agent dispatch.