subagent-driven-development

Extract all tasks from the plan into a todo list, then dispatch a fresh implementer subagent per task with the full task text and curated context. After each task, run a spec compliance review followed by a code quality review before marking it complete.

Subagent-driven execution runs each task in a fresh subagent with isolated context and automatic review checkpoints in the same session. Inline execution suits tightly coupled tasks where subagent isolation adds overhead without benefit.

Spec compliance confirms the code matches requirements with nothing missing or extra, which must hold before judging quality. Reviewing quality first wastes effort on code that may need rework to meet the spec.

Assess the blocker: provide more context and re-dispatch if it is a context problem, use a more capable model if it needs deeper reasoning, split the task if too large, or escalate to the human if the plan itself is wrong.

No. The Skill explicitly forbids parallel implementation subagents because they cause conflicts. Tasks run sequentially, each completing its full review cycle before the next begins.

Avoid it when tasks are tightly coupled and need shared evolving context, when no implementation plan exists yet, or when the coordinator has not confirmed subagent-driven execution mode with the user.