subagent-driven-development

Dispatch specialized subagents for each task—one for implementation, one for spec compliance review, and one for code quality review. This approach isolates context, prevents quality drift, and enforces structured verification across your development plan.

Use fresh subagent instances per task instead of maintaining a single agent across all work. Each subagent starts clean, reducing accumulated context drift and improving precision in both implementation and review stages.

Implement TodoWrite-based task tracking to log each subagent's work, decisions, and review outcomes. This creates an auditable record that ties implementation tasks to their spec and code-quality verification results.

Use this approach when your implementation plan has largely independent tasks requiring fast iteration and structured verification. It's most effective for complex projects where quality checks and spec compliance must be separated from coding work.

Yes. This Skill requires explicit reviewer prompts tailored to your spec and code standards. Each subagent needs clear instructions to enforce consistent compliance and catch quality issues before they accumulate across tasks.

Coordination overhead increases with task count, and subagent consistency depends on prompt clarity. This approach works best for medium-to-large plans with well-defined tasks; small, tightly coupled projects may not justify the added structure.