subagent-driven-development

Executing independent tasks from a single implementation plan requires dispatching a fresh subagent for each task, enabling isolated work and parallel progress within a single session. This approach applies task orchestration to activate each task individually.

A two-stage review workflow for plan execution applies spec compliance review followed by code quality review for every task. This structured process ensures prompt-driven validation through implementer execution and self-review before finalization.

Preventing context contamination when automating parallel tasks involves assigning a fresh subagent per task. Keeping context clean ensures isolated work and rapid iteration without cross-task contamination during multi-task plan execution.

Task orchestration works for plans with five parallel independent tasks by coordinating automatic review loops within a single running session. It assigns a new subagent for each task and manages reviews until all tasks are complete.

Automating prompt-driven workflows for complex plans is best achieved through subagent-driven development. This mechanism dispatches fresh subagents to handle per-task implementer execution and coordinates two-stage reviews until completion.

You should not use fresh subagents for task orchestration when your implementation plan lacks multiple independent tasks. This approach is designed specifically for multi-task plans requiring isolated work and two-stage reviews within a single session.