dispatching-parallel-agents

To debug multiple unrelated test failures simultaneously, you split independent problems into separate parallel agent tasks. This approach assigns one focused agent per failing area, allowing investigations to run concurrently without interfering with each other.

The best way to delegate independent implementation tasks is grouping unrelated work items by subsystem or file, then assigning each a focused agent with explicit constraints. This ensures isolated context and clear expected outputs for every concurrent task.

You should split debugging work into parallel agents when facing multiple test failures or workstreams that lack shared state or sequential dependencies. This allows simultaneous investigation of independent subsystems, avoiding slow serial debugging.

To coordinate and integrate findings after dispatching parallel agents, you review returned findings, reconcile changes across isolated contexts, and verify the combined result. This ensures the separately investigated subsystems merge correctly.

When assigning tasks to parallel agents for debugging, you need to define focused scopes, isolated context, explicit constraints, and clear expected outputs. This prevents independent investigations from interfering with each other.

Parallel task delegation fails with shared state because this approach requires independent problems without sequential dependencies. If subsystems share state, isolated agent contexts will conflict, preventing proper coordination and integration of combined results.