subagent-driven-development

Subagent-driven development dispatches a fresh, isolated subagent per task to execute plans, ensuring clean context and preventing cross-task leakage. It coordinates implementer and reviewer prompts to drive work and track progress.

The two-stage review process evaluates spec compliance first, then code quality, to ensure high-quality outcomes across tasks. This structured task orchestration coordinates implementer and reviewer prompts to drive work and re-dispatch as needed.

You can orchestrate parallel progress by providing a plan in the current session and dispatching a fresh subagent per task. The skill applies per-task TodoWrite tracking and built-in escalation to coordinate prompts and iterate until all tasks pass.

No external dependencies are required. The skill maintains clean context by dispatching a new isolated subagent per task directly within the current session, using structured task orchestration and prompt templates to drive implementation.

You should use isolated subagents when executing complex plans that require parallel progress and strict context isolation, preventing cross-task leakage. A two-stage review process ensures spec compliance and code quality across all dispatched tasks.

The best way to track progress is using per-task TodoWrite within the isolated subagent workflow. The skill coordinates implementer and reviewer prompts, applying built-in escalation and re-dispatching subagents until spec and quality reviews pass.