subagent-driven-development

Automated code review in a TDD workflow is handled by dispatching a fresh subagent per task, then enforcing a mandatory two-stage review loop that checks spec compliance and code quality with fix-and-recheck cycles until approval.

Subagent-driven development is a parallel task orchestration method that launches isolated subagents for independent implementation tasks, providing per-task context isolation to avoid context pollution and maintain clear accountability.

Spec compliance is enforced before code quality checks by running an automated two-stage review loop, where a spec compliance review executes first, followed by a code quality review, with fix-and-recheck cycles until both stages pass.

Subagent-driven development applies to implementation plans with mostly independent tasks. It dispatches fresh subagents per task, so plans with heavy task dependencies may not suit the per-task context isolation and parallel execution model.

Integration testing after parallel subagent implementations is performed by running a final integration review and full test suite verification before merge, ensuring all independently implemented tasks integrate cleanly and pass terminal and file toolset checks.

Fresh subagents for each task remove inconsistent, ad-hoc implementations by providing per-task context isolation, preventing context pollution across tasks, and maintaining clear accountability throughout the implementation plan.