parallel-agent-dispatch

Parallel agent dispatch assigns a dedicated agent to each independent failure domain to triage and propose fixes concurrently. This enforces focused agent scopes, prevents cross-domain edits, and outputs a consolidated summary of findings.

The best way to triage unrelated subsystem failures like auth and payments is to identify independent failure domains and run one agent per domain. This allows you to tackle issues in parallel without shared state, then merge changes into a combined report.

You should use parallel agents for issue triage when three or more failures exist, are unrelated, and can be tackled in parallel without shared state. This method isolates investigations and prevents cross-domain edits across different subsystems.

Parallel agent dispatch does not work for failures with shared state dependencies. It is specifically designed for investigating multiple independent failures that can be tackled in parallel without shared state to prevent cross-domain edits.

To consolidate findings after running parallel agents on different domains, the system outputs a consolidated summary of findings for integration. You dispatch tasks in parallel, review the combined report, and merge changes from each focused agent scope.