dispatching-parallel-agents

Dispatch parallel agents to investigate and fix independent failures concurrently. The Skill launches one agent per problem domain, verifies independence, monitors progress across all agents simultaneously, then aggregates and integrates results—drastically reducing resolution time versus sequential debugging.

Use parallel dispatch when you have 3 or more independent failures across separate domains with no shared state. Each failure must be investigable in isolation. This approach eliminates sequential bottlenecks; for example, fixing 5 unrelated failing test files in parallel saves significant time.

The Skill enforces independence verification as part of its structured process. It guides you to confirm that failures span disjoint domains, share no state, and can be resolved without conflicts. This prevents agent tasks from interfering with one another.

After all agents finish, the Skill performs post-dispatch conflict checks to detect unintended interactions, then runs full integration verification to ensure combined fixes work together correctly without regressions.

No. Parallel dispatch applies only to independent failures across disjoint domains with no shared state. Dependent or related failures require sequential or coordinated resolution, not concurrent dispatch.