subagent-driven-development

Parallel task execution is orchestrated by spawning fresh subagents for each independent task, enabling rapid iteration without cross-session planning. A dedicated subagent is dispatched per task to avoid context pollution and coordinate explicit completion signaling.

The two-stage review workflow enforces spec compliance first, followed by a code quality review. This process ensures tasks dispatched to subagents meet functional requirements before evaluating the structural integrity of the implementation.

You should use subagent-driven development when your tasks are mostly independent and require quick turnover. It is primarily useful for large plans containing independent tasks that need rapid iteration without cross-session planning.

No, fresh subagents prevent cross-session plan execution. The framework is designed strictly for local, session-scoped parallel execution to tackle independent tasks within the same active session.

Spawning a fresh subagent per task prevents context pollution across independent tasks. This ensures each subagent maintains a clean execution state, enabling rapid iteration and accurate task completion signaling.

The framework coordinates per-task subagents using explicit completion signaling. It dispatches dedicated implementer subagents for each independent task and reviews the results in two stages to maintain workflow consistency.