dispatching-parallel-agents

Debug concurrent failing test suites by dispatching one isolated agent per failing test file. This parallel debugging approach groups failures by subsystem so each agent owns a focused problem domain, allowing independent root causes to be remediated simultaneously without cross-contamination.

Isolated context execution is a parallel debugging mechanism where dispatched agents do not inherit shared session history. This prevents cross-contamination and coordination overhead, ensuring each agent independently investigates its assigned failures without interference from other concurrent investigations.

Yes, you can use parallel agent dispatch for race condition investigations when each issue can be understood without shared system state. The orchestrator spawns isolated agents that independently investigate concurrent failures across different subsystems, provided the failures have unrelated root causes.

Integrate fixes from parallel debugging agents safely through conflict-aware integration. After concurrent investigations complete, the orchestrator verifies that the isolated agents' changes do not conflict with each other, then validates the integrated codebase by re-running the full test suite.

You should not use parallel debugging when multiple test failures share the same root cause or require shared system state to understand. This approach requires independent-domain partitioning, meaning each failure must be isolated enough to be investigated without cross-agent coordination.