dispatching-parallel-agents

Group failures by independent problem domain, then dispatch one agent per domain with a focused prompt containing the failing test names, error messages, constraints, and expected output. After agents return, review summaries, check for conflicting edits, and run the full test suite.

Use parallel dispatch when failures are independent, such as different test files or subsystems with different root causes and no shared state. Avoid it when failures are related, when understanding requires full system context, or when agents would edit the same files.

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 such as a summary of root cause and fixes.

Common mistakes include overly broad scopes like fixing all tests at once, missing context such as error messages, absent constraints that let agents refactor unrelated code, and vague output requirements that hide what actually changed.

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