staff-subagent-driven-development

You can execute an implementation plan by dispatching a fresh subagent for each independent task and reviewing the result before continuation. This approach keeps your controller session clean and coordinated while ensuring isolated context for every delegated coding step.

Subagent delegation works by assigning each independent task from your implementation plan to a fresh subagent, which executes the work in isolated context. The controller session then reviews the subagent's result against the spec and code quality standards before moving to the next task.

The two-stage review process checks spec compliance first, then code quality, for each task a subagent completes. This ensures mistakes are caught before work advances, keeping your implementation plan on track and preventing context drift across sequential handoff points.

No, subagent-driven development applies to same-session software development workflows where tasks are executed sequentially. It dispatches a fresh subagent for each independent task and reviews the result before continuation, requiring explicit handoff points and completion tracking across the full plan.

Yes, task breakdown is required. Task orchestration supports implementation plans with multiple independent steps, question handling, and completion tracking, so you must decompose your feature work into discrete, independent tasks before dispatching them to fresh subagents.

The best way to manage code review for delegated tasks is a two-stage loop that checks spec compliance first, then code quality. This ensures each subagent's work meets requirements before the controller session advances to the next task in the implementation plan.