dispatching-parallel-agents

Debug multiple unrelated test failures by dispatching parallel agents to investigate each problem domain independently, preventing wasted time and unnecessary context switching. Each focused agent receives a self-contained prompt with clear scope, constraints, and expected outputs to resolve failures concurrently.

Coordinate independent engineering tasks by decomposing them into focused agent assignments that proceed concurrently without sequential dependencies. This workflow orchestration requires self-contained prompts, explicit constraints, parallel execution, conflict review, and full-suite verification before integration.

Use parallel agents for workflow orchestration when you identify failures, subsystems, or work items that have no shared state or sequential dependencies. This allows focused work to proceed concurrently, which is ideal for unrelated test files or independent debugging tasks.

Integrate changes from parallel agents by comparing their results, checking for conflicts, running verification, and integrating successful changes. You must perform a conflict review and full-suite verification before integration to ensure the parallel execution did not introduce regressions.

You cannot dispatch parallel agents for subsystem investigations with sequential dependencies or shared state. This approach applies strictly to independent engineering tasks that can proceed without shared state, requiring self-contained prompts and explicit constraints for each focused agent.

Create self-contained prompts for focused agent delegation by defining clear scope, constraints, context, and expected outputs for each specialized agent. This ensures the parallel execution remains isolated and produces verifiable results for conflict review and integration.