dispatching-parallel-agents

Group failures by independent problem domain, then dispatch one subagent per domain in the same response so they run concurrently. Give each agent a focused prompt with the failing test names, constraints, and a required summary of findings and fixes.

A good agent prompt is focused on one problem domain, self-contained with all needed context such as error messages and test names, explicit about constraints like not changing production code, and specific about the expected output summary.

Avoid parallel dispatch when failures are related and fixing one might fix others, when understanding requires full system context, during exploratory debugging where the cause is unknown, or when agents would share state and interfere with each other.

Issue all subagent dispatch calls in the same response to trigger parallel execution. Dispatching one agent per response results in sequential execution, which loses the concurrency benefit.

Review each agent's summary to understand the changes, check whether agents edited the same code and created conflicts, run the full test suite to verify all fixes work together, and spot-check results since agents can make systematic errors.