dispatching-parallel-agents

Group failures by root cause domain, then dispatch one subagent per independent file with a focused prompt containing the error messages, scope constraints, and required output. After agents return, review summaries, check for conflicts, and run the full test suite.

Use parallel dispatch when failures are independent—different test files, subsystems, or bugs with no shared state. If fixing one failure might fix others, or if understanding requires full system context, investigate sequentially with a single agent instead.

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

Avoid parallel dispatch when failures are related and one fix might cascade, during exploratory debugging where the broken domain is unknown, when full system context is needed, or when agents would share state and interfere with each other.

Read each agent's summary to understand what changed, verify the agents did not edit the same code, run the full test suite to confirm all fixes work together, and spot-check results since agents can make systematic errors.