subagent-driven-development

Subagent-driven development automates task execution by deploying a fresh subagent per task, then runs parallel code, business logic, and security reviews between steps. This eliminates context pollution and accelerates high-quality execution across independent tasks within the same session.

Yes. The Skill dispatches a fresh subagent per task and aggregates review issues by severity, inserting TODO/FIXME notes and committing changes before advancing to the next task. Parallel execution and automated review gates ensure fast iteration with quality control.

Load your plan into TodoWrite, dispatch a Task tool per independent task with instructions, then trigger parallel reviews by code-reviewer, business-logic-reviewer, and security-reviewer. The Skill aggregates findings, inserts notes, commits, and advances automatically.

Use fresh subagents per task when executing independent tasks requiring continuous progress and automated quality gates. This prevents context pollution, maintains focus, and enables true parallel execution—ideal for high-throughput development workflows.

Subagent-driven development is optimized for independent tasks within the same session where each task can be isolated. Dependent tasks requiring sequential handoffs are not the primary use case; the Skill excels at parallel, quality-gated execution.